Compare commits

...
197 Commits
Author SHA1 Message Date
rUv 6699a2fc44 Match README spatial SVGs to RuView header palette and add RF wave motion 2026-10-12 20:33:57 -07:00
rUv 964377765a Improve README SVG spatial motion and typography (#2257) 2026-10-12 20:28:12 -07:00
rUv 5cfbb8d98d docs: restore original GIF at top of README (#2252) 2026-10-12 14:30:42 -07:00
rUv 0f95193190 Merge pull request #2248 from ruvnet/codex/toolkit-release-20261012
fix(toolkit): integrate capture/device corrections for 0.9.2
2026-10-12 09:54:03 -07:00
rUv 15fd5eebdc Merge pull request #2246 from ruvnet/codex/sensing-release-20261012
fix(calibration): finish failed sessions and reconcile decision diagnostics
2026-10-12 09:27:10 -07:00
ruv b064acf57c test(harness): cover Node 20 WebSocket capability boundary 2026-10-12 12:20:40 -04:00
rUv 4413491370 Merge pull request #2247 from ruvnet/codex/firmware-release-20261012
fix(firmware): correct identity and secure OTA provisioning for 0.8.14
2026-10-12 09:17:31 -07:00
ruv b265454b93 fix(remote): isolate Windows PowerShell module search paths 2026-10-12 12:09:26 -04:00
ruv 99d9851173 test(devices): align end-to-end medians with even-sample correction 2026-10-12 12:01:15 -04:00
ruv 720cfd7c8a fix(remote): replace explicit Windows ACL grants with current-user access 2026-10-12 11:56:13 -04:00
ruv 2c1222789d Merge remote-tracking branch 'origin/main' into codex/toolkit-release-20261012 2026-10-12 11:49:52 -04:00
ruv 3620abff12 fix(toolkit): integrate capture and device corrections for 0.9.2
Integrate reviewed source PRs #2201-#2223, #2225-#2231, #2235-#2240, #2242 and #2243 (firmware README hunks remain in #2247). Regenerate one package manifest after integration, validate native Windows test execution, bound MTK listener startup, support official C6 release bundles, and retain stronger release metadata gates.
2026-10-12 11:49:52 -04:00
ruv a937a5fd20 fix(firmware): correct device identity and provision OTA keys for 0.8.14 2026-10-12 11:48:38 -04:00
ruv 9d863c3d22 fix(cli): report calibration persistence failures as aborted 2026-10-12 11:43:04 -04:00
mallesh-bot 99cea559f9 fix(cli): retain abort status after finalize failure 2026-10-12 11:39:41 -04:00
mallesh-bot 18d98d9452 fix(cli): abort incomplete calibration sessions 2026-10-12 11:39:40 -04:00
ruv 4eb991a101 Merge decision diagnostics with landed holdout explanations 2026-10-12 11:39:24 -04:00
rUv 1c435a0a79 Merge pull request #2190 from ruvnet/docs/cognitum-animated-visuals
docs: add Cognitum-style animated README visuals and linked capability cards
2026-10-12 08:38:54 -07:00
rUv 3c8f7ed96e Merge pull request #2189 from ruvnet/prototype/uncertainty-virtual-view
Add opt-in Dream Machine virtual-view research adapter
2026-10-12 08:38:50 -07:00
rUv 3b0b8d72df Merge pull request #2198 from ruvnet/codex/bootstrap-holdout-diagnostics-20261010
fix(sensing-server): explain bootstrap holdout failures
2026-10-12 08:38:07 -07:00
Dragan Spiridonov ce93244117 feat(sensing): integrate decision and holdout diagnostics for Mac qualification 2026-10-11 14:05:06 +02:00
ruv 8a140c1906 fix(docs): harden README visual XML and path validation 2026-10-10 13:41:26 -04:00
ruv d779961a55 fix(virtual-view): validate bounded plain data before broker calls 2026-10-10 13:29:51 -04:00
ruv f59fe9d7c2 fix(sensing-server): explain bootstrap holdout failures 2026-10-10 13:22:26 -04:00
rUv a71634ae8a Merge pull request #2192 from clonea1/contrib/pin-led-strip-idf54
fix(firmware): pin led_strip to 3.0.x; 3.1.0 breaks ESP-IDF 5.4 builds
2026-10-10 10:15:56 -07:00
JoeandClaude Opus 5.5 f1ff58afe7 fix(firmware): pin led_strip to 3.0.x; 3.1.0 breaks ESP-IDF 5.4 builds
led_strip 3.1.0~1 calls rmt_tx_switch_gpio (IDF >= 5.5 API), which is
undeclared on ESP-IDF 5.4 (src/led_strip_rmt_dev.c:189), so every
clean-checkout firmware build fails. Constrain to ~3.0.3.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 14:28:05 -04:00
rUv 4696804b23 docs: add Cognitum-style animated RuView visual system 2026-10-09 05:45:16 -07:00
rUv 88daeb8cf2 Add virtual-view prototype: integrations/virtual-view/README.md 2026-10-09 05:40:33 -07:00
rUv f8750a4a50 Add virtual-view prototype: integrations/virtual-view/adapter.test.mjs 2026-10-09 05:40:32 -07:00
rUv 2c7a0194f4 Add virtual-view prototype: integrations/virtual-view/adapter.mjs 2026-10-09 05:40:30 -07:00
rUv 0ef6b96fe1 docs: describe RuView-focused README animation 2026-10-06 09:25:36 -07:00
rUv 2d5472372f docs: focus animation on RuView sensing and spatial awareness 2026-10-06 09:25:34 -07:00
rUv efdf97ecc1 docs: add ruOS animation below RuView header 2026-10-06 09:12:38 -07:00
rUv 46d80ffc3a docs: upload Ruflo-powered ruOS desktop animation 2026-10-06 09:12:36 -07:00
rUv 0b15c0ebc9 Merge pull request #2163 from ruvnet/release/aepod-integration
release: integrate aepod PRs #2100-#2157 (38 PRs, review fixes, validated)
2026-10-05 13:52:42 -07:00
ruv 682a6f4538 docs(changelog): community batch PRs #2100-#2157
Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-05 12:21:07 -04:00
ruv a72174c571 fix: review findings and cross-PR integration fixes for aepod release set
- mediatek: cap per-device maps at 64 with oldest-first eviction; clamp --csi-ring
- training: mark live-buffer fallback runs degraded (data_source/degraded) instead of plain "completed"
- multistatic: skip CIR blend when any frame in the cohort has a placeholder phase
- groundtruth: median count baseline; radar labels carry evidence=CLAIMED
- spatial-evidence: default proof CODE; MEASURED requires a reproducer
- scripts: collect-ground-truth guards non-JSON replies and unreadable token files
- docker: entrypoint accepts 1/true/yes/on for RUVIEW_UDP_INSECURE_LAN
- adaptive model: untrack v2/data/adaptive_model.json (#2135)
- docs: ADR-379 evidence tags, RUNBOOK/README OTA wording, NOTICE for vendored MediaTek patch
- tests: fresh timestamps for privacy tests (#2119 staleness), plan_source Result handling,
  drop obsolete qualcomm-binds-nothing test (#2126), ignore slow live-binary test (#2157)

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-05 12:08:21 -04:00
ruv fd323e483e Merge PR #2128 (aepod) into release integration
# Conflicts:
#	.github/workflows/ci.yml
2026-10-05 11:48:40 -04:00
ruv 9f0821cf69 Merge PR #2104 (aepod) into release integration 2026-10-05 11:48:25 -04:00
ruv 69fc9efff7 Merge PR #2145 (aepod) into release integration 2026-10-05 11:48:25 -04:00
ruv 20386cf313 Merge PR #2141 (aepod) into release integration
# Conflicts:
#	docs/adr/README.md
2026-10-05 11:48:24 -04:00
ruv c4fd6d9b36 Merge PR #2101 (aepod) into release integration
# Conflicts:
#	v2/crates/wifi-densepose-sensing-server/src/training_api.rs
2026-10-05 11:48:16 -04:00
ruv e223d86ba2 Merge PR #2144 (aepod) into release integration 2026-10-05 11:47:35 -04:00
ruv 52e5ebefbd Merge PR #2143 (aepod) into release integration 2026-10-05 11:47:35 -04:00
ruv ca44cd1a7b Merge PR #2117 (aepod) into release integration
# Conflicts:
#	docs/integrations/home-assistant.md
#	v2/crates/wifi-densepose-sensing-server/src/main.rs
2026-10-05 11:47:34 -04:00
ruv 5fe0a6a48d Merge PR #2105 (aepod) into release integration
# Conflicts:
#	v2/crates/wifi-densepose-sensing-server/src/main.rs
2026-10-05 11:46:33 -04:00
ruv da1ef6a17d Merge PR #2102 (aepod) into release integration
# Conflicts:
#	docs/adr/README.md
2026-10-05 11:46:15 -04:00
ruv b6c2fb5b7d Merge PR #2116 (aepod) into release integration 2026-10-05 11:46:08 -04:00
ruv b8b70b7df7 Merge PR #2115 (aepod) into release integration 2026-10-05 11:46:07 -04:00
ruv bb5d903724 Merge PR #2114 (aepod) into release integration
# Conflicts:
#	v2/crates/wifi-densepose-sensing-server/src/main.rs
2026-10-05 11:46:07 -04:00
ruv 1e2348abef Merge PR #2113 (aepod) into release integration
# Conflicts:
#	v2/crates/wifi-densepose-sensing-server/src/main.rs
2026-10-05 11:45:52 -04:00
ruv 2493185ab8 Merge PR #2112 (aepod) into release integration 2026-10-05 11:45:33 -04:00
ruv 678f63f4f3 Merge PR #2125 (aepod) into release integration
# Conflicts:
#	v2/crates/wifi-densepose-sensing-server/src/cli.rs
#	v2/crates/wifi-densepose-sensing-server/src/main.rs
2026-10-05 11:45:26 -04:00
ruv 0aba0fc467 Merge PR #2127 (aepod) into release integration
# Conflicts:
#	v2/crates/wifi-densepose-sensing-server/src/main.rs
2026-10-05 11:44:53 -04:00
ruv 6f550c39ed Merge PR #2157 (aepod) into release integration 2026-10-05 11:44:34 -04:00
ruv e9d0cef610 Merge PR #2133 (aepod) into release integration 2026-10-05 11:44:34 -04:00
ruv bb22f1b185 Merge PR #2109 (aepod) into release integration 2026-10-05 11:44:33 -04:00
ruv 7bc5fe1a53 Merge PR #2103 (aepod) into release integration 2026-10-05 11:44:33 -04:00
ruv 51e1788c7e Merge PR #2129 (aepod) into release integration 2026-10-05 11:44:32 -04:00
ruv 7b62297f3e Merge PR #2126 (aepod) into release integration 2026-10-05 11:44:31 -04:00
ruv 890e521a3a Merge PR #2132 (aepod) into release integration 2026-10-05 11:44:31 -04:00
ruv 444184f9ef Merge PR #2119 (aepod) into release integration 2026-10-05 11:44:30 -04:00
ruv 7955af5bc0 Merge PR #2107 (aepod) into release integration 2026-10-05 11:44:29 -04:00
ruv 273a2a8536 Merge PR #2108 (aepod) into release integration 2026-10-05 11:44:29 -04:00
ruv add0755958 Merge PR #2134 (aepod) into release integration 2026-10-05 11:44:28 -04:00
ruv de5f7edd16 Merge PR #2100 (aepod) into release integration 2026-10-05 11:44:22 -04:00
ruv 69e0729c61 Merge PR #2156 (aepod) into release integration 2026-10-05 11:44:22 -04:00
ruv 9baaf8434f Merge PR #2155 (aepod) into release integration 2026-10-05 11:44:21 -04:00
ruv 27eefb6a7f Merge PR #2140 (aepod) into release integration 2026-10-05 11:44:20 -04:00
ruv c6f7e8ca0d Merge PR #2120 (aepod) into release integration 2026-10-05 11:44:20 -04:00
ruv 474f1a904e Merge PR #2118 (aepod) into release integration 2026-10-05 11:44:19 -04:00
ruv ddaa54eba0 Merge PR #2110 (aepod) into release integration 2026-10-05 11:44:19 -04:00
ruv b3d31510ee Merge PR #2130 (aepod) into release integration 2026-10-05 11:44:18 -04:00
ruv bce512cb8d Merge PR #2131 (aepod) into release integration 2026-10-05 11:44:18 -04:00
ruv 9950ecc340 fix(provision): refuse to write credentials through a symlink with a clear error
Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-05 11:44:17 -04:00
ruv e0c4207bd5 Merge PR #2152 (aepod) into release integration
# Conflicts:
#	firmware/esp32-csi-node/provision.py
2026-10-05 11:43:47 -04:00
ruv 044a41d2de Merge PR #2153 (aepod) into release integration 2026-10-05 11:43:14 -04:00
ruv 8d243a67f4 Merge PR #2111 (aepod) into release integration 2026-10-05 11:43:13 -04:00
Mathew Beaneandclaude-flow ee6717f035 fix(sensing-server): lock the CSI grid gate on the majority grid and report a stalled pipeline on /health (#1894)
The per-node grid gate locked onto the densest CSI grid a node had ever
sent and rejected every other grid. Real ESP32-S3 nodes interleave
several grids: one node sent 192 bins on 88% of frames and 306 on 6%, so
the gate locked onto the 306 minority and dropped about 90% of that
node's frames. A node whose radio moved to a sparser grid for good was
rejected forever; when every node did this, `tick` froze while /health
still reported ok.

The gate now votes over each node's last 64 frames and locks onto the
grid the node sends most. A challenger must outnumber the active grid
3:2 to take over, so interleaved grids don't flap the lock. A switch
clears the window as before. On an ESP32-C6 the HE-SU majority still
wins and the HT minority is still rejected.

/health keeps `status: "ok"` (the UI's server probe keys on it) and adds
a `processing` block: a live/stalled/no_input/idle state, the age of the
last published update, and lock-free UDP counters (datagrams, age of the
last datagram, datagrams dropped by the source allowlist). A frozen
pipeline can now be told apart from input that stopped arriving.
Allowlist drops also log a warning on the 1st, 2nd, 4th, 8th... drop
instead of only at debug level.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-04 17:31:47 -04:00
Mathew Beaneandclaude-flow f5a36666fc fix(hardware): accept ESP32 frames up to the firmware buffer size
Esp32CsiParser capped n_subcarriers at 256. Real ESP32-S3 nodes on
firmware 0.8.12 send 1-antenna frames with 306 subcarriers (612 I/Q
bytes), so the parser dropped about a third of live frames. Pointcloud
started hitting that cap once it switched to this parser.

The bound now comes from the firmware: frames are serialized into
CSI_MAX_FRAME_SIZE = 20 + 4 * 256 * 2 bytes, so a frame carries at most
1024 I/Q pairs across all antennas. The check is n_antennas *
n_subcarriers <= 1024.

HT/legacy bandwidth no longer maps subcarrier counts above 128 to 80 or
160 MHz. ESP32 radios use 20/40 MHz channels, and the larger counts
(192, 306) come from merged LLTF/HT-LTF buffers, so anything above 64
is reported as 40 MHz.

The pointcloud UDP receive buffer grows from 2048 to 4096 bytes so the
largest legal frame (2068 bytes) is not truncated.

Refs #1348

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-04 16:52:10 -04:00
Mathew Beaneandclaude-flow d78a221607 feat(provision): read the WiFi password from a file or a prompt
--password puts the WiFi password in argv, where ps and shell history can
see it (#1754). This adds two ways to give it without that:

- --password-file PATH reads it from a file, dropping one trailing newline
  (\n or \r\n). The file must exist and not be empty. If group or others can
  read it, the script refuses and suggests chmod 600.
  --allow-insecure-password-file accepts such a file with a warning, for
  read-only mounts such as 0444 secrets.
- When --ssid is given with no password flag and stdin is a terminal, the
  script asks with getpass. It reuses a saved password only if it was saved
  for the same SSID. Without a terminal (scripts, CI) nothing is asked and
  the existing missing-credentials check applies.

--password still works, but now prints a one-line warning on stderr that
points at --password-file. The two flags are mutually exclusive. The
firmware README examples now use --password-file.

Tests: 10 new tests in tests/test_provision_secrets.py, with placeholder
credentials and getpass and the flasher stubbed. 8 fail on the previous
commit; the other 2 check that nothing is asked without a terminal or when a
saved password matches the SSID. The suite goes from 25 to 35 passing.

Refs #1754

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-04 16:50:43 -04:00
Mathew Beaneandclaude-flow db81dce7ba fix(multistatic): gate phase combining on a declared shared reference
The sensing-server bridge builds every node frame with a zero phase
placeholder and the fuser combined those phases across nodes as if they
were measured against one reference. ESP32 nodes each run an independent
crystal with a random PLL phase after reset; mesh time sync aligns frames,
not RF phase, so there is nothing to combine.

- Add PhaseReference (Unavailable / NodeLocal / Shared{id}) to
  MultiBandCsiFrame. The server bridge declares Unavailable.
- MultistaticFuser combines phase only when every frame declares the same
  Shared id. Otherwise fused_phase is empty and FusedSensingFrame carries
  PhaseFusion::AmplitudeOnly(reason). Amplitude fusion is unchanged.
- The ADR-134 CIR blend is skipped when the phase is a placeholder.
- TrustedOutput and /health/ready (trust.phase_fusion) report the mode
  and reason.
- ADR-029 status note updated.

Refs #1752, #1710, #1750.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-04 16:42:03 -04:00
Mathew Beaneandclaude-flow 3a8ec8139e fix(pointcloud): parse ESP32 CSI frames with the reference ADR-018 parser
The pointcloud parser read channel from byte 8, RSSI from byte 9, noise
floor from byte 10 and a "timestamp" from bytes 16..19. The firmware
(csi_serialize_frame) writes freq_mhz as u32 at 8..11, sequence at
12..15, RSSI at 16, noise floor at 17 and PPDU type/flags at 18..19. So
RSSI came out as the high byte of the frequency (+9 for 2437 MHz), and
the tomography attenuation was built from it.

parse_adr018 now delegates to wifi_densepose_hardware::Esp32CsiParser,
the same parser the rest of the workspace uses, and only reshapes its
output. CsiFrame.channel/timestamp_us become freq_mhz/sequence, which
are what the header actually carries. ADR-081 feature-state packets
(0xC5110006) are no longer decoded as CSI frames.

build_test_frame (used by csi-test) now emits the real layout. Golden
byte-by-byte tests cover single and two-antenna frames, truncation and
the feature-state packet; a cross-check test compares against the
reference parser.

Refs #1348

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-04 16:30:45 -04:00
Mathew Beaneandclaude-flow 98c1d4a1e9 fix(provision): key saved state by board MAC, not serial port
provision.py saved each board's settings under the serial port path. Ports
get reused when boards are swapped (a second board on the same hub slot often
comes up as the same /dev/cu.usbmodem*), so the next board inherited the
previous board's node_id, target_port and channel without any sign (#1755).

- On a real flash the script reads the base MAC with `esptool read_mac` and
  keeps state in mac-<aabbccddeeff>.json. The record also stores the MAC and
  the port it was last written from.
- EUI-64 chips (C5, C6, H2) print an 8-byte "MAC:" line before "BASE MAC:";
  the parser takes the base MAC on those and the single MAC line elsewhere.
- Migration: if the board has no record yet and a port-keyed file from an
  earlier version exists, it is used once, with a notice that points at
  --reset, then removed after the save so no other board can pick it up.
- --state and --dry-run still never open the port. --state shows the board
  last provisioned from that port, and --mac selects a board by hand.
  --dry-run without --mac keeps the port-keyed file, which the QEMU swarm test
  relies on (--port /dev/null --dry-run).
- If the MAC can't be read, state stays keyed by port and the script says so.

Tests: tests/test_provision_mac_state.py adds 18 tests, using placeholder
MACs, with esptool, the NVS generator and the flasher stubbed. 15 of them
fail on the base; the other 3 guard behaviour the base already has (no serial
access on --dry-run, --reset, rejecting a bad --mac). The suite goes from 14
to 32 passing.

Not verified on hardware: the real esptool read_mac call. The output shapes
in the tests come from esptool 4.12 and 5.4.0 cmds.py.

Refs #1755

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-04 16:28:47 -04:00
Mathew Beaneandclaude-flow c00deadaf7 fix(provision): keep the WiFi password owner-only and out of --state output
The per-port state file holds the WiFi password and seed token in clear, but
save_state() created the directory and file with the default umask, so they
landed 0755/0644 and any local user could read them (#1754). --state also
printed both values verbatim, and that output gets pasted into issues.

- The state dir is created 0700 and state files 0600. Staging uses mkstemp
  (unique name, O_EXCL, 0600) plus fchmod, so a stale temp file or a planted
  symlink can't receive the secret.
- harden_state_dir() tightens the dir and every .json/.tmp file in it,
  including ones left by older versions. main() runs it before any read or
  early exit, so --state repairs old files too. It skips symlinks and files
  owned by other users, and only warns if chmod fails.
- The --dry-run NVS binary and the generator-failure CSV fallback are written
  0600 even when the file already exists. The generator's CSV and binary now
  live in a private mkdtemp() dir instead of the shared temp dir.
- --state shows password and seed_token as (set)/(empty). --show-secrets
  prints them.

Windows is unchanged: chmod can't set owner-only ACLs there.

Tests: tests/test_provision_secrets.py adds 11 tests; all 11 fail on the
base and the suite goes from 14 to 25 passing.

Refs #1754

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-04 16:25:43 -04:00
Dragan Spiridonov 7c8aeacea2 Merge pull request #2147 from ruvnet/codex/calibrated-vitals-empty-20261004
fix(sensing): gate vital outputs on fresh calibrated occupancy
2026-10-04 15:07:58 +02:00
Dragan Spiridonov c3764a429f fix(sensing): require calibrated occupancy before publishing vitals 2026-10-04 11:50:13 +02:00
Mathew Beaneandclaude-flow 985e5d9449 feat(groundtruth): radar + CSI occupancy auto-labeller with session split and baseline
auto_label cuts a RuView .csi.jsonl recording at a radar log's window
boundaries and labels each cut empty/occupied with the radar's person
count and count interval, plus a small CSI summary (frame count, mean
RSSI, mean and variance of per-frame amplitude). The radar-to-recording
clock offset is a required, reported parameter. Windows are dropped with
a counted reason when the radar window is UNKNOWN, when too few CSI
frames fall inside it, or when an "empty" comes within empty_hold_ms of
an occupied window (a motion tracker can lose someone who sits still).

session_disjoint_split splits by session only and refuses a split where
either side would be empty or a named test session has no windows.
majority_baseline fits the no-CSI model on the training side (majority
occupancy class, mean count) and scores it on the test side, so any
occupancy accuracy built on these labels has a baseline next to it.

examples/radar_autolabel.rs is a small CLI over the three. Tests use
synthetic data only; nothing here has been run on a real capture yet.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-03 23:55:11 -04:00
Mathew Beaneandclaude-flow b033fd309e feat(groundtruth): radar window adapter for presence, count and range labels
ruview-groundtruth had an MmWave reference modality but no way to read a
radar log. RadarLabels reads one JSON summary per window (the shape a
UART radar reader writes: timestamp_ms, window_start_ms, health, source
flags, presence, target_count, max_target_count, distance_cm, targets
x/y) and turns each window into presence, person-count and nearest-range
labels.

Each label carries its basis and an uncertainty: a count interval from
the counts the window reports, the configured range sigma (a datasheet
figure, so CLAIMED), and the reader's quality. Windows from a simulated
or unverified source, a no_source reader, or with a presence flag that
disagrees with the count are UNKNOWN with a reason rather than guessed.
Labelled windows convert to ReferenceSeries for the existing alignment
and agreement code. Malformed lines are row-numbered InvalidRow errors.

serde_json moves from a dev-dependency to a dependency for the parser.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-03 23:55:10 -04:00
Mathew Beaneandclaude-flow 76af08a9b0 docs(adr): ADR-381 node self-survey from pairwise ranging
Proposed. Defines the RangeObservation record, a warn-only check mode
against --node-positions with tolerances from MEASURED ESP32-S3 FTM
bench data, per-boot calibration, and a fill mode gated off until a
source shows MEASURED p90 link error <= 0.3 m (expected from UWB).
Records the FTM bias between responder starts plainly and parks FTM
for check mode, coarse gating and time sync rather than dropping it.
Links ADR-144, ADR-281, rufield ADR-261, #2109 and #2133.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-03 23:08:04 -04:00
Mathew Beaneandclaude-flow 951ffcc254 feat(survey): add check-mode node self-survey crate (ADR-381)
Adds ruview-survey, a host-side check that compares pairwise node ranges
with the configured --node-positions and warns, never writes. It reports
per-link residuals, a fitted scale with feet/inch/cm/mm unit detection,
a suggested id swap (exhaustive up to 8 nodes, fewest-changes rule,
explicit ambiguity for symmetric layouts), moved nodes, and unknown or
unranged ids.

RangeObservation is source-agnostic (FTM, UWB, BLE CS, manual). Per-boot
calibration keys offsets by (initiator, responder, responder_boot)
because the measured FTM bias moved with every responder start. Fill
mode is not implemented; fill_gate keeps it off until a source has
MEASURED p90 link error <= 0.3 m over >= 6 links.

FTM tolerances come from a two-board ESP32-S3 bench (20 MHz, ch 1):
spread within a start <= 0.31 m, bias between starts -0.5 to +1.9 m at
1.14 m, clamp to 0 below ~0.8 m. Tests are SYNTHETIC.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-03 23:08:04 -04:00
Mathew Beaneandclaude-flow cbd84acf6c fix(unified): stop GaussianMap::decay compounding across calls
decay(now) multiplied confidence by exp(-(now - timestamp_ns)/tau_eff)
on every call, but timestamp_ns is the last-seen time and does not move
when decay runs. Calling decay(t1) then decay(t2) applied
exp(-(t1-ts)/tau) * exp(-(t2-ts)/tau) instead of exp(-(t2-ts)/tau), so
confidence fell faster the more often the caller ticked.

Track how far each Gaussian has already been decayed (decayed_to_ns)
and only apply the interval since max(decayed_to_ns, timestamp_ns).
A fresh observation still restarts the clock, and the lifetime-stretched
tau_eff is unchanged. Fusion and the merge pass keep the later of the
two decayed_to_ns values. The new field is #[serde(default)], so maps
serialized before this change load with 0 and decay from timestamp_ns
as before.

Adds two tests: one-shot vs. per-second decay must agree to 1e-12, and
the wall-vs-transient persistence case must hold when decay runs every
10 s.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-03 22:58:13 -04:00
Mathew Beaneandclaude-flow a1d7eb0bfd docs(spatial-evidence): ADR-382 for the RF wire contract, pin the v1 yaw convention
The spatial.evidence.v1 format is specified in WeftOS ADR-107 section 7.
ADR-382 records the part RuView emits (envelope, room_enu frame,
rf_gaussian, rf_link_observation, validation ranges, versioning) so the
contract is readable from this repository, and names ADR-107 as its origin.

ADR-107 was amended on 2026-10-03: one yaw convention (degrees
counter-clockwise from room +x), new tof_depth and radar_range types, and
an optional sensor block on radar_track_point. The two RF bodies did not
change. The crate now documents the yaw sense of an rf_gaussian
orientation and tests it (+90 degrees about +z maps local x onto room +y),
and tests that the amended example lines are rejected as unknown types by
this RF-only mirror.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-03 22:57:08 -04:00
Mathew Beaneandclaude-flow 3741e7ba43 feat(spatial-evidence): emit RF Gaussians and links as spatial.evidence.v1
New leaf crate ruview-spatial-evidence converts ruview-unified output
into the spatial.evidence.v1 JSONL wire format (WeftOS ADR-107 section 7):

- RfGaussian / GaussianMap -> rf_gaussian records. A Gaussian with
  synthetic provenance is always tagged SYNTHETIC; the caller can lower
  the proof tag, never raise it.
- Link measurements (tx/rx, measured vs free-space amplitude) ->
  rf_link_observation with excess_loss_db = 20 log10(free / measured).
  The Friis reference comes from ruview-unified's channel model.

The v1 envelope and the two RF bodies are mirrored with local serde
structs, so there is no WeftOS dependency. Validation follows the v1
rules: schema version, id charset and length, finite values, coordinate,
frequency, scale, occupancy, confidence and excess-loss ranges, and the
16 KiB line limit. Golden tests compare against the ADR example lines.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-03 18:18:41 -04:00
Mathew Beaneandclaude-flow fabb0c8233 fix(sensing-server): refuse a routable HTTP bind when API auth is off
The UDP receiver already refuses a routable bind without a source
allowlist (ADR-296), and the Docker entrypoint refuses a routable bind
without RUVIEW_API_TOKEN (#864). Running the binary directly with
--bind-addr 0.0.0.0 and no auth only logged that auth was off.

The server now exits at startup when the HTTP/WebSocket bind is not
loopback and API auth (RUVIEW_API_TOKEN or RUVIEW_OAUTH_ISSUER) is off,
unless RUVIEW_ALLOW_UNAUTHENTICATED is set, in which case it starts and
logs a warning. Loopback binds are unaffected. The conditions match the
entrypoint's, so a container that starts today still starts.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:26:26 -04:00
Mathew Beaneandclaude-flow da2304985a fix(sensing): stop absent nodes reporting a person count of 1
Every person count is already gated on presence, but two presence
decisions disagreed with the motion level published next to them:

- smooth_and_classify[_node] set presence = smoothed_motion > 0.03 while
  the debounced level only leaves "absent" above 0.04. A quiet room that
  settles in (0.03, 0.04] read motion_level "absent" with presence true,
  so the count stayed at 1.
- The raw-CSI per-node count was gated on the per-frame classification,
  which an adaptive-model label overrides. /api/v1/nodes could show a
  node as "absent" with person_count 1 while the ADR-297 room vote (which
  reads the node's debounced level) already said the room was empty.

Presence now follows the debounced level, as the room vote already does,
and the per-node count is gated on the node's own published level.
score_to_person_count keeps its floor of 1: it only runs when presence is
true, and an empty room reports 0 through the gate.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:21:48 -04:00
Mathew Beaneandclaude-flow 8c6d29bf83 fix(sensing-server): report measured stream fps and real WebSocket client count
/api/v1/stream/status returned "fps": 10 whenever tick > 1, whatever the
real rate, and both it and /health counted every broadcast subscriber
(recorder, MQTT bridge) as a client. The dashboard Streaming card read
"IDLE, 0 client(s)" for up to 30 s while the page was streaming, because
its first /health poll lands before the page's socket connects; with auth
on, the ticket round-trip widens that gap.

- New stream_stats module: UpdateRateMeter (sliding 5 s window, bounded at
  1024 samples) and WsClientCounter with an RAII guard.
- AppStateInner::advance_tick() replaces the five `s.tick += 1` sites and
  records each sensing update in the meter.
- /api/v1/stream/status now returns the measured update rate as `fps`
  (0 when nothing flows), `active` from that rate, `clients` from the
  WebSocket counter, `nodes` with each fresh node's measured csi_fps (null
  until its EMA has 5 samples) and `csi_fps_total`.
- /health and the /health/health stream component count WebSocket clients
  on /ws/sensing and /api/v1/stream/pose; the stream component also carries
  the measured rate.
- DashboardTab re-reads health 500 ms after the sensing socket connects or
  drops, so the card follows the socket instead of waiting for the next
  30 s poll.

Fixes #2087

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:15:23 -04:00
Mathew Beaneandclaude-flow 5cffe9e7db fix(sensing-server): accept vendor --source values and bind UDP for them
The previous commit limited --source to auto/esp32/wifi/simulated, which
rejected the vendor names the server already keys on: mediatek, qualcomm,
realtek (RTL8720F radar) and realtek_csi (RTL8721Dx CSI). They are the
labels the UDP receiver writes and effective_source() ages.

On 03a89793 those names fell into the catch-all plan, which bound no UDP
receiver, so vendor frames were never ingested. They now plan like esp32
(UDP receiver bound) without the ESP32-shaped simulator. Anything else,
including case variants and runtime-only labels like "mediatek:simulated",
is still rejected.

Refs #2097

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:13:38 -04:00
Mathew Beaneandclaude-flow c51a1cf888 fix(sensing-server): apply --privacy-mode to REST, WebSocket and recordings
--privacy-mode was declared in the MQTT arg group and read only by the
MQTT publisher, so /api/v1/sensing/latest, /api/v1/pose/current, the
/ws/sensing and /api/v1/stream/pose streams and recordings kept serving
pose keypoints (and heart/breathing rate when vitals are published).

- Move --privacy-mode (env RUVIEW_PRIVACY_MODE) out of MqttArgs to a
  top-level server flag. Spelling and env var are unchanged. The MQTT
  config takes its value from the server flag.
- Add privacy_filter: one key filter for vital_signs, HR/BR leaves,
  enhanced_breathing, keypoints, pose_keypoints and joints_m.
- REST: middleware strips those keys from every JSON response on the
  HTTP and WS ports. Unparseable JSON is withheld with a 500.
- WebSocket: /ws/sensing frames and /api/v1/stream/pose messages are
  filtered; the pose physics assessment is withheld.
- /api/v1/pose/current?view=both|refined returns 403 privacy_mode.
- Recordings write filtered frames.
- Default (flag off) behaviour is unchanged; the middleware is a no-op.

Fixes #2094

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:12:45 -04:00
Mathew Beaneandclaude-flow 522374584c fix(ui): stop retrying and ask for the API token when the server requires one
With RUVIEW_API_TOKEN set on the server and no token in the browser, the
sensing socket was refused, closed with 1006, and retried forever while the
header said "Connecting...". A browser cannot see the 401 on a refused
upgrade, so the service could not tell this from an outage.

- ws-ticket.js: resolveWsUrl() reports a stored token the server refuses
  (401/403) as 'rejected'. probeWsAuthRequired() repeats the ticket request
  with no bearer after an unticketed socket fails to open; a 401 means the
  server wants a token this browser does not have ('missing'). The probe
  only runs on the failure path, so auth-off servers see no extra request.
  withWsTicket() keeps its string contract for the other callers.
- sensing.service.js: new 'auth-required' state and dataSource. Entering it
  clears the retry timer; reconnect() leaves it.
- The header widget shows "Token required", the global banner explains it,
  and both open QuickSettings on API Access (once automatically per page
  load, then on click). Tab banners and the dashboard indicator get labels.
- Tests: ws-ticket.test.mjs (+6) and a new sensing.service.test.mjs (5),
  added to the CI UI test step.

Fixes #2099

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:12:23 -04:00
Mathew Beaneandclaude-flow 5755d9f11f fix(sensing-server): status waits for frames; reject unknown --source
/api/v1/status reported source_state "live_unverified" for a configured
live source before any frame had arrived, because source_state() fed a
fixed zero frame age into the ADR-295 resolver. It now uses the real age
of the last accepted frame for the current source, so the states are:

- disconnected (waiting_for_frames: true) until the first frame
- live_unverified while frames arrive within ESP32_OFFLINE_TIMEOUT (5 s)
- stale once they stop for longer than that

The response also carries waiting_for_frames and last_frame_age_ms.
ADR-270 vendor events get their own receive clock so they age the same
way.

An unknown --source value (for example "macos") used to start a server
with no data task. Clap now rejects it with the list of valid values
(auto, esp32, wifi, simulated, simulate), and plan_source returns an
error for anything else.

Docs: user guide lists the valid --source values and status states, the
macOS section uses --source wifi, and the Linux section says there is no
Linux RSSI source in the server yet.

Refs #2097

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:09:50 -04:00
Mathew Beaneandclaude-flow 3bf993275e fix(sensing-server): give the governed fuser the configured node positions
EngineBridge built its StreamingEngine without ever calling
set_node_positions, so the governed fuser placed every node at the
origin even when --node-positions was set. Startup now passes the same
keyed map the sibling multistatic_fuser gets, via
EngineBridge::with_node_positions. The fuser resolves positions by
node_id per frame, so nodes discovered later need no extra call.

/api/v1/status (and /health/ready) now reports the map under
trust.node_positions so the configuration can be checked at runtime.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:08:58 -04:00
Mathew Beaneandclaude-flow 05ea4feaed fix(sensing-server): keep recordings and models under --data-dir
Recordings, the recording list/delete routes, the trainer's dataset
loader and model export, the models routes and the adaptive classifier
all used paths relative to the process cwd (`data/recordings`,
`data/models`, `data/adaptive_model.json`), so `--data-dir` only covered
config and the session secret.

Add `recordings_dir(data_dir)` and `models_dir(data_dir)` and use them
everywhere. `MODELS_DIR` still overrides the models directory, and the
trainer now honours it too. With the default `--data-dir data` the paths
are unchanged.

Fixes #2088

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:07:11 -04:00
Mathew Beaneandclaude-flow 7af5477f04 fix(scripts): parse the ADR-018 header layout in the UDP capture scripts
Both capture scripts decoded a header layout the firmware does not send.
csi_collector.c and esp32_parser.rs use: magic u32 @0, node_id @4,
n_antennas @5, n_subcarriers u16 @6, freq_mhz u32 @8, sequence u32 @12,
rssi i8 @16, noise_floor i8 @17, ppdu_type @18, flags @19, then
n_antennas * n_subcarriers int8 I/Q pairs from @20.

- collect-training-data.py read a 16-byte header and int16 I/Q, so its
  length check rejected every real single-antenna frame. It now reads the
  20-byte header and int8 pairs, records freq/channel/seq/ppdu/flags, and
  lets the recorder stamp receive time (the header has no timestamp).
- record-csi-udp.py read rssi from byte 6 and channel from byte 7 (the
  n_subcarriers bytes) and treated bytes 8-19 as I/Q, so a 64-subcarrier
  frame came out as 70 subcarriers with header bytes in the first six
  amplitudes. It now checks the magic, reads the real fields, and writes
  only the I/Q block to iq_hex.
- Add scripts/tests/test_capture_adr018.py, which builds frames with the
  esp32_parser.rs test layout and checks the decoded fields.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 22:01:22 -04:00
Mathew Beaneandclaude-flow 84f5e19e9d fix(scripts): run the proof with python3 and report INCOMPLETE when a gate is skipped
prove.sh looked only for `python`, so on macOS (python3 only) it skipped
the deterministic proof and still printed RESULT: PASS with exit 0.

- Pick $PYTHON, then python3, then python.
- Check numpy/scipy import before running verify.py and print the
  lockfile install command when they are missing; warn when installed
  versions differ from archive/v1/requirements-lock.txt.
- Run verify.py single-threaded, as CI does.
- Hard gates that cannot run (no cargo, no Python, missing deps) are
  reported as [NOT RUN] and the run ends INCOMPLETE with exit 2.
  Data/hardware-gated claims are unchanged.
- Tell people to stop any local server on UDP 5005 first (#2086).

Fixes #2095

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 21:57:49 -04:00
Mathew Beaneandclaude-flow 1a50162598 fix(sensing-server): stop serving a frozen /sensing/latest as current
latest_update is only rewritten when a source produces a new update. When
every source went quiet, /api/v1/sensing/latest kept returning the last
classification indefinitely, with no marker that it was old. A one-off
bogus frame could leave "absent 0.97" frozen there.

Past 10 s, the endpoint now returns
{"status":"stale","age_ms":...,"source":effective_source()} instead of
the old payload. Fresh payloads gain an age_ms field. "no data yet" is
unchanged. This is separate from #2060, which excludes stale nodes from
the room vote: that changes what a new update says, while this bounds how
long an old update is served.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 19:22:26 -04:00
Mathew Beaneandclaude-flow 0eb952a9c5 fix(sensing-server): evict stale node_states on a timer, not only on frame arrival
Stale node eviction (60 s) lived inside the ESP32 receive path and ran
every 100 ticks, so it only happened while ESP32 frames kept arriving.
After a burst of bogus node ids, or once every node went quiet, nothing
ever removed the entries, and they stayed listed in /api/v1/nodes for the
life of the process.

Move the existing rule unchanged into evict_stale_node_states (60 s,
nodes that never reported included). Call it from the receive path as
before, and from a new node_states_eviction_task that sweeps every 30 s
alongside the UDP receiver.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 19:22:26 -04:00
Mathew Beaneandclaude-flow 08f0f8859b fix(sensing-server): reject ESP32 CSI frames with zero antennas or subcarriers
parse_esp32_frame accepted any datagram that started with the 0xC511_0001
magic and had a 20-byte header. With n_antennas or n_subcarriers at zero,
the expected length collapses to the header itself, so the length check
passed and the frame minted a node carrying empty amplitude/phase arrays.
A non-ESP32 sender on the shared UDP port produced exactly that: bogus
nodes in /api/v1/nodes.

Return None when either count is zero. Well-formed frames parse as before.
The new tests cover zero antennas, zero subcarriers, and a bare header,
and check that a well-formed frame still parses.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 19:22:26 -04:00
Mathew Beaneandclaude-flow 8c63d7b558 docs(mqtt): document the published entity set and add an ADR-115 P4.5 status note
home-assistant.md now describes current behaviour: the 15 announced
entities (13 with --privacy-mode) with their source and availability
rule, the entities that are not announced and why, publish rates and
retention, the persisted client id, and what privacy mode does and does
not cover. The guide no longer documents --semantic* flags that
sensing-server does not accept, and it drops the HIPAA/GDPR/CCPA
claims.

ADR-115 §11 records what P4.5 wired and what it did not, with a
before/after measurement against a real broker (MEASURED, one live
node).

Refs #2085, #2093

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 18:46:21 -04:00
Mathew Beaneandclaude-flow e351d0fce1 feat(mqtt): wire semantic states into the publisher, announce only sourced entities
Discovery announced 20 entities per device, but only 6 ever received
state (#2085). presence/state was published on every CSI frame, and the
per-process default client id gave Home Assistant new devices on every
restart (#2093).

- mqtt::planner maps each snapshot to availability, state and events.
  It runs one SemanticBus per node at 1 Hz (ADR-115 P4.5) and gates
  binary states on change with a 60 s retained heartbeat. Sensors keep
  their configured rates. Each entity is `online` only while the server
  has a source for it; otherwise it is `offline` and gets no state.
- Entities with no source on the sensing broadcast are no longer
  announced: zone_occupancy, bathroom_occupied, bed_exit,
  multi_room_transition (no zones), meeting_in_progress (motion is
  three-level) and pose (pose_keypoints is never set). 15 remain, 13
  with --privacy-mode.
- mqtt::bridge, moved from main.rs, reads only sensing_update. That
  removes the phantom aggregate device that edge_vitals frames created.
  It takes fall events from the edge_vitals rising edge and
  motion_energy from motion_band_power (previously always 0.0), and maps
  present_still to 0 % motion.
- The default client id is wifi-densepose-<8 hex>, persisted in
  <data-dir>/mqtt_client_id. --mqtt-client-id still overrides it.

Refs #2085, #2093

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 18:46:11 -04:00
Mathew Beaneandclaude-flow ea6da967d2 docs(mediatek): OpenWrt WN586X3 CSI patch set, router CSI guide, ADR-266 amendment
- firmware/openwrt-wn586x3/: the patch set that made a Wavlink WL-WN586X3
  Rev A (MT7981B + MT7976C) emit CSI on OpenWrt 24.10.8. It contains:
  MediaTek's mt7915 CSI driver patch rebased onto the mt76 revision 24.10.8
  pins (BSD-3-Clause-Clear, MediaTek author line kept); the Rev A board
  patch (GPL-2.0); a buffer-overrun fix for `mt76-vendor dump csi`; and
  an attribute-enum fix for MtkCSIdump. Each patch keeps the licence of
  the code it modifies. There are no images, firmware blobs, or captures.
- docs/mediatek-router-csi.md: an operator guide covering supported
  silicon, the image build route that was tested, arming CSI, the
  bridge, `--source mediatek`, endpoints and provenance labels, what the
  data is not, evidence, and troubleshooting. Every claim is tagged
  MEASURED, CODE-DERIVED, CLAIMED, or SYNTHETIC.
- scripts/mediatek-csi-phase-stability.py: a read-only reproducer for
  the one MEASURED offline result. On one MT7976C radio, the cross-chain
  phase arg(H0*conj(H1)) moves 0.014-0.039 rad (p50) between adjacent
  frames for well-sampled transmitters, against about 1.56 rad for raw
  per-chain phase, which is what random phase gives. It never prints a
  transmitter address. The captures it was run on are not in the tree.
- ADR-266 Amendment 1: vendor CSI exists on MT7981 through MediaTek's
  public feed patch, demonstrated on one board model, and still
  `mediatek:physical-unvalidated`. The hardware gates remain open.
- user-guide.md and the bridge README link to the guide.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 18:32:25 -04:00
Mathew Beaneandclaude-flow f44f25d596 feat(mtk-bridge): bridge MT7981 vendor CSI into MTC1
Adds `wifi-densepose-mtk-bridge`. It converts MediaTek mt76 vendor CSI
into ADR-267 MTC1 frames for the sensing server's UDP ingest.

Inputs:
- `--replay`: an `mt76-vendor dump csi` JSON file, or a `.mtkcap`
  recording. Records are grouped into frames by the driver's last-chain
  bit, per transmitter address, so HT 2x2 and legacy 1x2 runs from one
  radio each close with their real dimensions. Concatenated and duplicate
  dumps (the vendor tool appends) are reported, not hidden.
- `--listen`: MtkCSIdump UDP datagrams. They carry no packet identity, so
  each datagram becomes one 1x1 frame rather than a guessed pairing.
  Missing RSSI/noise is the -128 dBm sentinel, and PPDU type is `Unknown`.

Provenance:
- The bridge never sets CALIBRATED.
- A replayed file must be either attested (`--captured-on
  <model>/<firmware>`) or declared `--synthetic`, or the bridge refuses to
  start.
- Synthetic is sticky. It can come from a sidecar, a `.synthetic.` file
  name, or a capture header, and none of these can be overridden.
- `--listen` requires a `--listen-allow` CIDR list. Other sources are
  dropped and counted, never decoded.

Fixtures are fabricated and use locally administered addresses. The
assembler tests that model the two live dump loops use stand-in addresses,
not the real clients'. Captures written under the bridge's earlier format
tag still load.

Tests: 51 unit and 39 integration tests, plus clippy -D warnings.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 18:30:00 -04:00
Mathew Beaneandclaude-flow 8ecac18978 feat(sensing-server): labelled MediaTek amplitude heuristic and activity index
Real MediaTek frames now produce a room update, but from a heuristic that
is labelled as one on every output. The validated ESP32 pipeline needs
per-subcarrier phase history this path does not feed it, so it is not
used here and is never impersonated.

- `mediatek-amplitude-heuristic-v0` (CLAIMED, no labelled evaluation):
  presence and motion from per-device mean and peak amplitude over a 10 s
  time window (not a sample count, because arrivals are bursty).
  Hysteresis means one dip cannot flip the room to absent. Below a
  confidence of 0.3, a device reports `unknown` and abstains from the
  room vote rather than asserting "absent".
  `SensingUpdate.classifier` names it; `nodes[].device_id` and
  `nodes[].mediatek_diagnostics` say why each device reads as it does.
- `activity-index-v0` (CLAIMED): a fast/slow fluctuation index against
  each device's own baseline, plus an absolute level against a learned
  quiet floor. It is served in `SensingUpdate.activity` and
  `GET /api/v1/csi/mediatek/activity`, with
  `is_occupancy_estimate: false` on every block. `--activity-state <path>`
  persists learned floors across restarts.
- The summary endpoint adds the heuristic's window statistics.
- The MQTT per-node topics key on `device_id` when it is present, so two
  64-bit receiver ids cannot collide on `node_id` 0.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 18:28:26 -04:00
Mathew Beaneandclaude-flow e5f88b88fe feat(sensing-server): per-device MediaTek CSI ingest and inspection
The server kept one MediaTek snapshot in total, so two MT7981 receivers
overwrote each other, and `--source mediatek` fell through to the
catch-all source plan and bound no UDP socket at all.

- `--source mediatek` binds the UDP receiver and does not start the
  ESP32-shaped simulator, which has no MediaTek branch and would otherwise
  publish a simulated room classification next to real frames.
- Snapshots are kept per `device_id`.
  `GET /api/v1/csi/mediatek/devices` lists them with `age_ms` and drops
  any device silent for over 30 s.
- A bounded per-device ring (`--csi-ring`, default 256 frames, about
  0.5 MiB for a 2x2 device at 64 subcarriers) keeps per-chain,
  per-subcarrier amplitude and phase. It serves
  `/devices/:device_id/frames` and `/devices/:device_id/summary`. Chains
  are keyed by (tx, rx) identity, so a device that interleaves 1x2 legacy
  and 2x2 HT frames is reported correctly.
- New provenance label `mediatek:physical-unvalidated` for real-silicon
  frames that are not calibrated. `SYNTHETIC` still takes precedence over
  every other flag. Plain `mediatek` stays reserved for calibrated frames.

No classifier changes in this commit.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 18:27:19 -04:00
Mathew Beaneandclaude-flow ebb303cbb8 feat(hardware): carry legacy and unknown PPDU types in MediaTek MTC1
A real MT7981 (WL-WN586X3, MT7981B + MT7976C) CSI report arrived with
rx_mode = MT_PHY_TYPE_OFDM for a 2.4 GHz client. The MTC1 PpduType had no
pre-HT value, so a bridge had to drop the frame. MediaTek's CSI driver
treats legacy OFDM as its own tone-mask group, and association traffic is
legacy-rate, so these frames are routine rather than a corner case.

Add PpduType::Legacy = 6 and PpduType::Unknown = 7. Values 1-5 and the
72-byte header are unchanged. A decoder built before this change rejects
6 and 7 as UnknownPpduType (fails safe, drops the frame), so MTC1
consumers need a rebuild.

ADR-267 gains Amendment 1 recording the change, its compatibility, and the
observed 1x1x64 dimensions of the motivating capture. The capture is not
in the tree: it holds a client transmitter address and raw CSI.

Developed in whitsentry, a RuView-based project.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 18:17:36 -04:00
Mathew Beaneandclaude-flow c80f618119 fix(sensing-server): show which nodes use the default position (#1804)
Nodes not named in --node-positions report [2.0, 0.0, 1.5], and the boot
line counts parsed entries whether or not any matches a live node id. A
config keyed by the wrong ids looked applied and did nothing.

- WARN once per node id when a fresh node has no configured position,
  naming the configured ids (INFO when no positions are configured).
- /api/v1/nodes: add position_configured per node.
- /api/v1/status: add node_positions with configured_node_ids,
  default_position and default_position_node_ids.
- At boot, WARN that entries without a node_id: prefix are keyed by list
  index from 0, with the entry -> node mapping. Semantics are unchanged.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 16:17:25 -04:00
Mathew Beaneandclaude-flow 3b542f8a85 fix(docker): make UDP CSI ingest posture explicit in the container
Since ADR-296 the sensing server binds its UDP receiver to 127.0.0.1 by
default, and docker-entrypoint.sh never set a UDP bind. Inside a
container that address cannot receive from the published 5005/udp port,
so the default image never shows a node (#2089, follow-up to #1394).

The entrypoint now binds UDP on 0.0.0.0 only when the operator has set
an ADR-296 source guard (RUVIEW_UDP_ALLOW, RUVIEW_UDP_INSECURE_LAN, or
the matching CLI flags). Without a guard it keeps the loopback bind and
prints a note explaining why no node will appear and what to set,
including the macOS bridge-gateway allowlist. An explicit RUVIEW_UDP_BIND
or --udp-bind still wins.

I chose this over defaulting the bind to 0.0.0.0 unconditionally. With
no guard that would make the server refuse to start (exit 1) in auto
mode, which binds UDP, and the guard would still have to be supplied. The
change keeps ADR-296's fail-closed rule as is: a routable bind never
happens without an allowlist or the explicit insecure opt-in. Auth
defaults are unchanged.

The entrypoint also normalizes RUVIEW_UDP_INSECURE_LAN=1/yes/on to true,
since clap rejects 1 for that flag.

The compose service passes RUVIEW_API_TOKEN, the RUVIEW_UDP_* variables
and SENSING_ALLOWED_HOSTS through from the shell, with notes on the
exit-64 auth check, the macOS gateway allowlist and the 421 Host check.
The README and user-guide Docker quickstarts now set a token, and the
user guide gains a "Receiving ESP32 frames in Docker" section.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 16:13:10 -04:00
Mathew Beaneandclaude-flow ef01c9e2be fix(sensing-server): accept 1/0/yes/no/on/off for boolean env flags
clap's default bool parser only takes true/false, so
RUVIEW_UDP_INSECURE_LAN=1 exited 2 with "invalid value '1'". Add a
shared parse_env_bool value parser (1/0, true/false, yes/no, on/off,
case-insensitive; anything else is still an error) and apply it to every
bool+env flag: RUVIEW_UDP_INSECURE_LAN, RUVIEW_NO_MDNS,
RUVIEW_NO_EDGE_REGISTRY, RUVIEW_MQTT, RUVIEW_MQTT_TLS,
RUVIEW_PRIVACY_MODE and RUVIEW_MATTER. The bare --flag form is
unchanged.

The Docker entrypoint now accepts RUVIEW_ALLOW_UNAUTHENTICATED=
true/yes/on (any case) as well as 1.

Fixes #2091

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 16:07:20 -04:00
Mathew Beaneandclaude-flow abeca0f0ed fix(sensing-server): send multi-node test frames to an ephemeral port
The multi_node_test UDP smoke tests sent synthetic ADR-018 CSI and
vitals frames to 127.0.0.1:5005, the sensing server's default port.
Because the tests are not ignored, a routine `cargo test` (and
scripts/prove.sh) could inject up to 255 fake node IDs into a locally
running server, and the hardcoded port could collide with other
processes.

Each test now binds its own receiver on 127.0.0.1:0 and sends to the
OS-assigned address, keeping the receiver alive for the test's
duration. The tests still only exercise the sender and frame-builder
path; no server is involved.

Ported from whitsentry, a RuView-based project, which carries the same fix.

Closes #2086

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 16:04:54 -04:00
Mathew Beaneandclaude-flow 5a36be17d2 fix(provision): make --dry-run and failed --reset leave state intact
--dry-run persisted the merged state record, so the next real run merged
on top of values that never reached the device. It now skips save_state.
The NVS binary is still written to nvs_provision.bin because ADR-061 and
scripts/qemu-mesh-test.sh consume it.

--reset deleted the per-port state file before argument validation, so a
run that then failed (e.g. missing WiFi credentials) still lost the record.
It now only ignores the prior state; a successful flash overwrites the file
as before.

Also fix the module docstring's package name: the PyPI package is
esp-idf-nvs-partition-gen, matching the runtime hint.

Fixes #2096

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 16:03:30 -04:00
Mathew Beaneandclaude-flow 559528155f fix(sensing-server): train the adaptive classifier per node, as it runs
Runtime classification is per node, from that node's own features and
amplitudes. The trainer instead built one sample per recorded line from the
fused frame-level features plus the first node's amplitudes, so a model
trained on a multi-node recording learned a different input distribution
from the one it is later asked to classify.

- load_recording now emits one sample per node per new frame, from that
  node's node_features entry and amplitude vector. A node contributes only
  when its amplitude vector changed since the previous line (every broadcast
  snapshots all nodes), and stale nodes are skipped. Lines without
  node_features (older single-node recordings) fall back to the frame-level
  features.
- Runtime subcarrier statistics use the same 56-subcarrier window that
  sensing_update, and therefore every recording, stores
  (RECORDED_AMPLITUDE_LEN, now shared with the broadcast in main.rs).
- The trainer stamps model version 2 (TRAINED_MODEL_VERSION), so a freshly
  trained model consumes the measured temporal dominant frequency (ADR-356)
  instead of being switched to the legacy subcarrier proxy at runtime.
- The training accuracy log line is labelled in-sample and printed as a
  percentage (it printed the 0..1 fraction with a % sign).

Tests: cargo test -p wifi-densepose-sensing-server --no-default-features:
bin 337 passed (5 new node_sample_tests), lib 539 passed,
integration tests pass. Clippy reports nothing on the changed lines.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:15:05 -04:00
Mathew Beaneandclaude-flow f161abf582 feat(sensing-server): loopback UDP tee for second consumers (ADR-380)
Add --udp-tee / RUVIEW_UDP_TEE: every datagram the source allowlist admits
is copied unchanged to up to four loopback targets, so a fusion host or
Cognitum cog can read full I/Q, sequence numbers and sync packets without
binding the CSI port. Targets must be loopback and may not reuse the
listener port or port 0. Sends are non-blocking and drops are counted. Off
by default.

Tests: cargo test -p wifi-densepose-sensing-server --no-default-features
--lib udp_tee (6 passed) and --bin sensing-server (332 passed). Loopback
smoke test of the debug binary: a 288-byte datagram sent to the UDP port
arrived byte for byte at the tee target, and a non-loopback --udp-tee
target made the server exit 1 with "must be a loopback address".

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:12:19 -04:00
Mathew Beaneandclaude-flow a4f80e63a2 docs(adr): ADR-379 OTA first-boot health check, with hardware evidence
Records the decision behind the two preceding firmware commits: why an
OTA'd image has to confirm itself, the health signals (STA IP plus one CSI
frame accepted by the stream sender), the 30 s minimum uptime and 120 s
deadline, the FORCE_FAIL test hook, ota_state in /ota/status, and the httpd
stack sizing.

MEASURED 2026-09-29:
- ESP32-S3 (16 MB, rollback bootloader): OTA push to ota_1, HTTP 200;
  "httpd stack after POST /ota: 4900 of 8192 bytes never used";
  pending_verify then valid; still valid on ota_1 after a hard reset.
  A FORCE_FAIL build pushed to ota_0 rolled back to ota_1 after ~30 s.
- ESP32-C6 (4 MB layout, rollback bootloader): push to ota_1, HTTP 200;
  pending_verify then valid ~28 s later; still valid after a hard reset.

Not yet verified: the 120 s no-IP timeout path, a true power cycle (an RTS
hard reset was used), and forced rollback on the C6.

Numbered 379: main has up to ADR-378 and open PRs claim nothing higher (2026-10-02).

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:12:12 -04:00
Mathew Beaneandclaude-flow 43f5f99108 fix(scripts): send the API token from collect-ground-truth and check the reply
With authentication enabled, collect-ground-truth.py could not start a CSI
recording: it sent no bearer token. It also treated any HTTP 200 as success,
but the recording endpoints answer 200 with {"success": false, "error": ...}
for refusals such as "recording already in progress". Either way it went on
capturing camera keypoints with no paired CSI and only printed a warning.

- Read the token from --token-file, else RUVIEW_API_TOKEN, and send it as
  Authorization: Bearer on start and stop.
- post_json() returns False when the body says success: false.
- Name the CSI recording gt_<timestamp> so it pairs with
  keypoints_<timestamp>.jsonl.
- Exit with an error when the recording cannot start, unless
  --allow-no-csi is given.

Tests: py_compile; scripts/tests (15 passed); a local smoke test against a
stub HTTP server checked the bearer header, the request id, the
success:false refusal, and token precedence (file, then env).

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:11:08 -04:00
Mathew Beaneandclaude-flow f8ab204711 fix(wifiscan): keep the redacted macOS connected link; auto+wifi keeps UDP bound
On macOS 27, without Location Services, CoreWLAN redacts both the SSID and
the BSSID, while rssi, noise and channel stay real. resolve_bssid abstained
when both identities were blank, so every macOS scan was dropped and
`--source wifi` silently produced nothing.

- mac_wifi --scan-once marks its sample `"connected": true`. It is always the
  single connected link, so a synthetic id cannot merge two networks.
- A redacted connected sample gets a stable locally administered synthetic
  BSSID; the SSID stays empty. Unmarked redacted lines still abstain.
- `--source auto` that resolves to host WiFi now also binds UDP, and wifi_task
  yields while the source is a live esp32, so ESP32 nodes that start after
  the boot probe still promote the source (the #1004 rule). Without this, the
  fix would make stock Macs ignore late-starting nodes.
- A missing helper is reported once, with the build command.
- ADR-025: Updated row + Amendment 1 recording the measured behaviour.

Measured on macOS 27: --source wifi gives source wifi:unnamed with sensing
ticks advancing; --source auto on a quiet port starts on wifi with
udp_receiver=true and promotes to esp32 (5 nodes) when real frames arrive.
Tests: wifiscan 158+1, sensing-server bin 332, all passing.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:11:08 -04:00
Mathew Beaneandclaude-flow 0df5069831 fix(training): train from real sensing recordings, per node
training_api loaded `{id}.csi.jsonl` in a flat {subcarriers, rssi} schema,
but the recorder writes `{id}.jsonl` sensing_update lines (nodes[] amplitude,
node_features[]). Training on a real recording therefore saw zero frames
(#1703).

- Accept sensing_update lines as well as legacy flat frames; try `{id}.jsonl`
  first, then `{id}.csi.jsonl`.
- Emit one frame per node per update, skip repeated snapshots and nodes
  without amplitudes, and keep frames grouped by node (stable sort).
- RecordedFrame gains an optional node_id; temporal features (window, prev)
  only use frames from the same node.
- Malformed lines are counted, not fatal.

Tests: recording_format_tests (4 new);
cargo test -p wifi-densepose-sensing-server --no-default-features
--bin sensing-server: 336 passed, 0 failed.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:11:07 -04:00
Mathew Beaneandclaude-flow f6a3a3579f fix(firmware): confirm OTA'd images with a bounded health check
Rollback-enabled nodes (CONFIG_BOOTLOADER_APP_ROLLBACK_ENABLE=y) boot an
OTA'd image as PENDING_VERIFY, but nothing called
esp_ota_mark_app_valid_cancel_rollback(), so every OTA reverted at the next
reset, a second OTA was refused until then, and a bad image was never
deliberately rolled back.

- main/ota_health.{c,h}: at boot, if the running image is PENDING_VERIFY,
  poll for STA got-IP and >=1 CSI frame accepted by the stream sender.
  Confirm after CONFIG_OTA_HEALTH_MIN_UPTIME_S (30 s) once both are in;
  roll back via esp_ota_mark_app_invalid_rollback_and_reboot() if they are
  not in by CONFIG_OTA_HEALTH_TIMEOUT_S (120 s from boot). VALID, UNDEFINED,
  NEW and no-otadata/factory images are left alone. The decision is a pure
  header function, host-tested in test/test_ota_health.c (make host_tests).
- CONFIG_OTA_HEALTH_FORCE_FAIL (test only) rolls back where it would have
  confirmed, for proving rollback on hardware.
- GET /ota/status reports ota_state so durability is visible remotely.
- RUNBOOK: replace the nonexistent ota_rollback_boot_check() claim and
  document the behaviour and the hardware check (2.1).

Hardware-verified on an ESP32-S3 (push, confirm, hard reset, forced
rollback) and an ESP32-C6 (push, confirm, hard reset); see ADR-379, next in
this series. Host tests: make host_tests passes, including run_ota_health.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:11:07 -04:00
Mathew Beaneandclaude-flow d25fb8192e fix(firmware): stop POST /ota overflowing the httpd stack
MEASURED 2026-09-29 on an ESP32-S3 (v0.8.12, 16 MB): every POST /ota of a
valid 1,228,672 B image died with "stack overflow in task httpd" and
RTC_SW_CPU_RST, so no OTA had ever completed on this firmware.

- ota_update.c: set config.stack_size = CONFIG_OTA_HTTPD_STACK_SIZE (new
  Kconfig, default 8192; HTTPD_DEFAULT_CONFIG gave 4096). Move the 1 KB
  receive buffer off the stack into a static (handlers are serialised on
  the one httpd task). Integer progress percent instead of %f.
  recv_wait_timeout stays 30 s.
- Log the httpd stack high-water mark after every POST /ota and
  POST /wasm/upload so the size rests on device data.
- wasm_upload.c: GET /wasm/list builds its 2 KB JSON on the heap.
- RUNBOOK §3 (over the air) records the failure and the high-water-mark
  log line.

Device check after the fix (same S3): "httpd stack after POST /ota: 4900 of
8192 bytes never used", i.e. about 3.3 KB peak. The ADR with the full
hardware evidence follows in this series (ADR-379).

This overlaps PR #1594, which raises the same stack. If #1594 lands first,
drop this commit; the rest of the series does not depend on it.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-02 03:11:07 -04:00
rUv 03a897938f Merge pull request #2062 from ruvnet/claude/ruview-npm-package-9uc5x3
feat(npm): RuView npm toolkit @ruvnet/ruview 0.9.1 — kernel, SDK, flashing, device access, MCP Apps, live Claude Code views (ADR-368..378)
2026-10-01 20:53:26 -04:00
ruv 15d48d4481 style(homecore-hap): rustfmt the new verifier test
Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 19:55:28 -04:00
ruv c764c8916b fix(homecore-hap): pad the SRP verifier to 384 bytes (flaky provisioning)
compute_verifier returns the minimal big-endian encoding, so about one
provisioning in 256 produced a verifier with a zero top byte, encoded to
383 bytes, and failed the store's own 384-byte check
("SRP verifier must contain exactly 384 bytes"). Seen in CI on
first_provisioning_returns_code_but_restart_does_not. Left-pad to the
3072-bit group width at creation: the same number, the fixed length.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 19:55:05 -04:00
rUv 5efc7e2c2c Merge pull request #2073 from ruvnet/feat/ruview-mcp-apps-ui
feat(npm): live sensor views in Claude Code, MCP Apps console, ESPHome radar, ruview umbrella — @ruvnet/ruview 0.9.1 (ADR-375..378)
2026-10-01 19:23:17 -04:00
ruv 39b4ae9ae1 release: @ruvnet/ruview 0.9.1 and ruview umbrella 0.9.1
Bump the harness and the umbrella (which pins it), point README and the
user guide at 0.9.1, sync the stale lockfile header (0.5.1, old bin name)
with package.json, and add a plain-language changelog entry for the
live sensing views in Claude Code (ADR-377, ADR-378).

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 18:57:49 -04:00
ruv 4a472926e1 fix(ruview-live): code-review findings on the waterfall (ADR-378)
- Stale frames shown as live: spectrumWith now keeps only nodes in the
  latest capture, so a node that stops streaming (or a failed capture)
  leaves the waterfall and the "no CSI frames" card returns.
- Multi-antenna ESP32 nodes never reached the waterfall: with --spectrum
  the parser adds chain-0 amplitudes (ADR-018 antenna-major) in a separate
  field the SpectrumCollector reads; the kernel path is unchanged.
- Frames from different CSI layouts could be joined: spectrum results
  carry their shape, and frames merge only when shape and bins match.
- Animation ran on assumed time: frames read the real clock (the one the
  full render uses), so pulses keep the reported rates and the replay
  keeps up when a frame fires late; replay lag is pruned with the nodes.

Tests: regression tests for each; npm test 193 pass, test:security 91
pass, engine tests 4 pass, validate passes.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 18:47:18 -04:00
ruv c4f9d95e11 fix(ruview-live): open as a dialog that takes the keys, sized per view
Hotkeys reach a pane only while it holds the keyboard, and `focus` alone
was not taking it. The engine's documented dialog form does: open with
focus + closeOnEscape + holdToasts (Escape closes, as originally). Each
open also requests `rows` for its view (overview 16, waterfall 30,
radar 26) so the inline pane opens tall enough for the bordered design
instead of a third of the screen; switching views re-requests the size,
and picture heights leave room for the chrome (rows - 15).

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 18:35:19 -04:00
ruv bf21c8c7a9 Revert "fix(ruview-live): fit the inline pane; actions on top; show focus state"
The compact inline layout read as broken next to the bordered cards the
user prefers; restore the previous design (78a147405). This reverts
commit 629e1b9c6.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 18:33:28 -04:00
ruv 629e1b9c6c fix(ruview-live): fit the inline pane; actions on top; show focus state
Outside fullscreen the pane sits inline above the prompt with a few rows,
so the bordered cards, rule and footer were cut off (Refresh/Close never
visible). Inline (or under 18 rows) the pane is now compact: views and
actions share the top row, no borders/rule/gaps, pictures sized to the
rows left, radar readings beside a small fan. The top row always says
whether the pane holds the keyboard ("keys active · Esc returns them" or
"click the pane or press ctrl+x tab to use the keys").

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 18:32:12 -04:00
ruv 78a1474057 fix(ruview-live): pane hotkeys need the keyboard — ask for focus, say how
Button hotkeys reach a Pane only while it holds the keyboard; the pane
opened without it, so 1/2/3/r/c went to the prompt. Open (and
/ruview waterfall|radar) now request `focus: true`; closeOnEscape is
dropped so Escape hands the keys back without closing; an unfocused pane
shows "click the pane or press ctrl+x tab to use the keys".

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 18:27:52 -04:00
ruv 751b7ae17b feat(ruview-live): showcase views — CSI waterfall, radar fan, animated cells (ADR-378)
The /ruview pane gains three views (keys 1/2/3) drawn in terminal Rasters
and animated in place with $.ui.blit (~12 fps, no redraws):

- CSI waterfall: per-subcarrier amplitude over time from the new
  `esp32 --spectrum` capture, two frames per row via the half block,
  5-95th percentile colour limits, replayed at the frames' arrival rate,
  labelled MEASURED or SYNTHETIC from the packets' own flag; n cycles nodes.
- Radar: a 120-degree range fan with rings, a fading distance trail and a
  sonar ping to the measured range; braille heart/breathing/distance charts
  scaled to the data; a heart and breathing gauge that pulse at the
  device-reported rates ("a metronome, not a waveform").
- A shimmer rule under the title; surfaces without Raster get a note.

Harness: captureEsp32 / ruview_esp32_capture take spectrum,
spectrum_bins (8-128) and spectrum_frames (8-256): a bounded ring of
binned amplitude frames per node, dominant shape, rate and synthetic flag.

The mod is split into model/raster/anim/views/register (own-file imports
only). Tests: 11 showcase + 11 mod Node tests, a --spectrum loopback test,
4 engine tests (waterfall tab sends --spectrum, draws the Raster, blits);
npm test 189 pass, test:security 90 pass, manifest verified, validate passes.

Evidence: MEASURED on an ESP32-C6 (COM11, release c6-adr110 v0.7.0, node 6)
streaming 7-18 Hz CSI to this host; a 1x256 capture renders the occupied
band against guard/null tones. Unpacked budget 384 -> 448 KiB (+~37 KB).

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 18:22:40 -04:00
ruv c4d7ef8086 feat(ruview-live): redesigned pane — LIVE badge, cards, trends, plausibility flags
- Header: RuView + LIVE/STALE badge (fresh within 2 refresh periods),
  "updated <time> (Ns ago)", a 5 s redraw tick keeps the age current.
- Bordered cards for the radar and the CSI nodes, side by side from 100
  columns, stacked below.
- Radar: PRESENCE DETECTED (+ target count) in green; distance, heart and
  breathing with per-refresh sparklines; device-reported vitals outside
  resting ranges (heart 40-180, breathing 4-40 bpm) are flagged in yellow
  instead of shown as fact (the live kit reported breathing 1.0 bpm).
- Nodes: per-node rate sparkline; "no packets" is a short hint in the card
  rather than a long top alert (status no longer counts it as an alert);
  an unconfigured radar says how to configure it.
- Buttons show their hotkeys.

Tests: 11 Node unit tests, 3 engine tests (mounted pane validates on the
terminal surface), validate passes.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 17:50:09 -04:00
ruv bd7c506076 fix(ruview-live): a reloaded pane waited forever for its first capture
A hot reload (or a resumed session) re-runs register with fresh variables
while the engine keeps the pane open, so no timer ran and the pane stayed
on "waiting for the first capture". Drawing the pane now resumes polling
when no timer is running. Engine test mounts the pane with no /ruview in
the module instance: fails without the fix, passes with it.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 17:44:00 -04:00
ruv 8e12606c6a fix(ruview-live): pane showed "updated Invalid Date"
$.clock.now() resolves a promise of epoch milliseconds; refresh passed the
promise into the model. Await it, and have the pane print "—" for any
non-finite timestamp. Seen live in Claude Code 2.1.287 (/ruview pane).
Regression tests fail without the fix (2/2) and pass with it.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 17:40:59 -04:00
ruv 8b857e41cd docs: README and user guide for @ruvnet/ruview 0.9.0 (devices, radar, MCP/ChatGPT, Claude Code mod)
- README MetaHarness section pinned to 0.9.0 with the hardware, MCP-over-HTTP,
  mod, Homecore and kernel commands; plugin section lists the ruview-live mod.
- User guide: new "RuView npm toolkit" section (hardware CLI, alert meanings,
  ESPHome kits, MCP/ChatGPT connection, the /ruview mod, Homecore + kernel),
  TOC entry, 0.5.0 pins updated.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 17:26:40 -04:00
ruv 093e3bbeae feat(ruview): ruview-live Claude Code mod shipped in the npm package (ADR-377)
- mod/: a Claude Code mod (function hooks). /ruview opens a pane beside
  the transcript with CSI nodes (rate, loss, RSSI, shape, SYNTHETIC) and an
  optional ESPHome radar (presence, distance, device-reported vitals),
  alerts (heartbeat-only boards + buffer-starvation hint), Refresh/Close
  buttons and a one-line status; refreshed on $.clock.every.
- Reads only through the harness CLI beside it (fixed read-only argv,
  --json) located via $.plugin.root (mods may import nothing else).
- Packaged in @ruvnet/ruview (mod/.claude-plugin, mod/hooks), listed in the
  repo marketplace as ruview-live@ruview; `ruview mod` prints how to load it.
- @ruvnet/ruview 0.9.0; umbrella ruview 0.9.0 (still unpublished).

Verified: claude plugin validate (mod + marketplace); claude plugin test
2/2 x3 runs; harness 172 tests + 89 security; real Claude Code headless
`/ruview refresh` from an installed tarball and with the live radar kit.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 17:12:53 -04:00
ruv 4128661789 docs(adr-376): record release outcome — components published, unscoped ruview refused by npm (403, similar to iview)
Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 16:45:20 -04:00
ruv b6f3abd35e feat(npm): ruview umbrella package + ESPHome radar source (ADR-376, ADR-373 A3)
- harness/ruview-umbrella -> npm `ruview` 0.8.0: one CLI and one MCP server
  for the operator harness, Homecore and the vitals kernel. `ruview homecore
  ...`, `ruview kernel <doctor|info|selftest|parity|synth|analyze|bench>`,
  `ruview capabilities`; everything else delegates to @ruvnet/ruview. MCP
  serves ruview_* + homecore_* tools and the ui:// console (stdio + HTTP).
- @ruvnet/ruview 0.8.0: bin renamed `ruview-harness` (the umbrella owns
  `ruview`; a clean tarball install showed the bin collision). New exports
  (cli, cli-ui, mcp, mcp-http, ui, esphome, kernel); injectable MCP handler;
  kernel as optional peer + setKernelImporter so the kernel resolves from
  the umbrella's tree under linked/pnpm installs.
- mmwave --source esphome: dependency-free read-only ESPHome native API
  client (private hosts only; encrypted/password APIs reported). Live Seeed
  MR60BHA2 kit: presence, distance, device-reported HR/BR (MEASURED).
- monitor detects Ameba CSI buffer starvation; radar views in terminal UI
  and console widget.
- homecore 0.1.0 and @ruvnet/ruview-kernel 0.1.0 gain ./cli exports.

Gates: harness 166 + 89 security, homecore 27 + 23, kernel 21 + 12,
umbrella 4; clean-dir tarball install smoke incl. live radar over HTTP MCP.
Local publish is a one-off exception to ADR-265 D2, recorded in ADR-376.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 16:30:23 -04:00
ruv 45796e4e3e feat(ruview): MCP Apps console widget, HTTP transport, terminal UI (ADR-375)
- ui://ruview/console-v1.html: one self-contained widget for MCP Apps hosts
  and ChatGPT (ui.resourceUri + openai/outputTemplate). Renders node
  streams, device scans and doctor results; Refresh re-runs the tool via
  window.openai.callTool or an MCP Apps tools/call. Empty CSP allowlist, no
  network, textContent only, parent-frame messages only.
- MCP: protocol negotiation (2025-06-18/2025-03-26/2024-11-05),
  structuredContent + compact JSON text (17-33% smaller, MEASURED),
  resources/list|read, tool titles, canonical tool name in result _meta.
  Transport-neutral handleRpc + shared FIFO dispatcher.
- `mcp start --http`: Streamable HTTP for ChatGPT/remote clients. Bearer or
  secret-path token (constant-time), loopback default, Origin allowlist,
  256 KiB cap, no batches, write grants never honoured over HTTP.
- CLI: formatted TTY output (tables, rate bars, alerts first, NO_COLOR),
  JSON for pipes/--json, capture countdown, `esp32 --watch` with per-node
  sparklines, lazy module loading (--help ~20% faster, MEASURED).
- monitor detects Ameba CSI report-buffer starvation ("lack of csi buf"),
  the root cause of the Realtek heartbeat-only stall seen live.
- Size budget 320 -> 384 KiB (package ~339 kB).

Verified: 160 tests + 83 security tests; headless-Chrome e2e of the MCP
Apps flow against live boards (refresh round trip 108 -> 105 packets);
HTTP security probes; clean tarball smoke (stdio + HTTP).

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 15:52:26 -04:00
ruv 3cd360f9e2 fix(ruview): CSI loss survives stray and reordered sequence numbers
Live RTL8721Dx capture (node 3, COM10) reported 97% loss: one stray
sequence value (+52,835, then straight back) was counted as 52,834 lost
frames, and every late UDP packet re-counted its gap. The tracker now
treats a single out-of-window value as a stray, needs two agreeing frames
to accept a counter reset, and lets a late packet fill its own gap
(seqReordered / seqStrays / seqResyncs in the summary). Against ground
truth on the live board (unique sequence numbers over the span): 4,783 of
4,783 present, loss 0, 9 strays; the tracker matched both.

ADR-373 Amendment 2 records the live Realtek run: devices -> realtek,
7,404/7,404 packets decoded, monitor at 1500000 baud without a reset, and
12,549 live frames analyzed by the WASM kernel.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 15:20:21 -04:00
Claudeandclaude-flow d31d9daba2 fix(nightly-sota): ignore stdin EPIPE from children that never read input
runProcess() ends the child's stdin with no error handler. A child that
never reads stdin (git check-ignore) can exit and close the pipe before the
write flushes; the unhandled EPIPE then crashes `agent.mjs validate`. This
failed harness/ruview (node 22) on PR #2062 and reproduces 12/12 locally
under CPU load (0/12 after this change). The command's real outcome is
still reported by the 'close' handler; this mirrors the existing handler in
harness/ruview/src/process-runner.js.

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_015H55jmprfz4oHWXSFbX5rf
2026-10-01 19:07:40 +00:00
ruv b25729acd6 fix(ruview): boot-log capture aborted on non-ASCII firmware logs on Windows
The flash flow's boot-log capture decoded serial lines and wrote text to a
stdout that the scrubbed child env leaves on cp1252. The ESP32-C6 release
firmware logs U+2192, so the capture died mid-boot (bootLog.captured=false);
an earlier abort would have hidden the CSI evidence and reported a good
flash as unvalidated. Write raw bytes instead. Regression test forces
PYTHONIOENCODING=cp1252 through a stand-in serial module (fails without
the fix). ADR-373 Amendment 2 records the real C6 flash evidence.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 15:02:09 -04:00
ruv 8d2be98ed7 feat(ruview): Realtek RAC1 nodes, honest capture, live CSI analysis (ADR-373 Amendment 2)
Findings from a hardware run of this branch against an RTL8721Dx board
(PL2303GC) and an ESP32-C6 (CP210x) on Windows 11:

- esp32 capture decodes Realtek RAC1 (ADR-323, CRC-verified, 8/16-bit
  tones) and RHB1 heartbeats, ADR-110 sync and ADR-081 mesh packets, and
  reports every CSI shape a node sends.
- Capture fails (no_decodable_packets / heartbeat_only) instead of
  returning ok:true + MEASURED when nothing decodes; names
  heartbeat-only senders.
- --analyze runs live single-antenna CSI through @ruvnet/ruview-kernel at
  the measured frame rate with a bounded buffer.
- monitor opens the port with DTR/RTS deasserted (no node reboot), takes
  --baud, and explains the C6/S3 USB-Serial/JTAG console.
- devices classifies Prolific 067B:23A3 as a Realtek board.
- Windows: owner-only ACL on the hosts file (fails closed); symlink test
  skips only where Windows refuses symlinks.
- kernel build names uninitialized submodules when cargo fails.
- Parser ~3.2x faster (MEASURED, 0.42-0.52M -> 1.33-1.67M packets/s).

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-10-01 14:40:32 -04:00
Claudeandclaude-flow 43f16e5638 perf(npm): binary kernel transport, faster device e2e, and CI fixes (ADR-368/373 amendments)
CI fixes (both failures on PR #2062 were in this branch):
- harness/ruview: remove `2.11.0` literals from training remedies; the
  ADR-265 D3 version-literal gate rejects them.
- ruview-kernel workflow: install wasm32 for the toolchain pinned by
  v2/rust-toolchain (the action only added it to `stable`); build.mjs now
  fails fast with the exact `rustup target add` remedy.

Kernel optimization (ADR-368 amendment):
- New `analyze_flat` / `session_push_flat` ops via `call_f64`: WASM
  `rvk_call_f64` (alignment-agnostic bytes) and napi `callF64(Float64Array)`.
  `analyze()`/`session.push()` pack uniform frames automatically; irregular
  input keeps the JSON path and its per-frame errors. `analyzeFlat()` is the
  zero-copy entry. Shared `Session::step` makes both paths byte-identical on
  identical f64 inputs (Rust test), and the binary path avoids serde_json's
  non-correctly-rounded float parsing.
- MEASURED (`ruview-kernel bench --seconds 120`, one Linux x64 host):
  2.82x (wasm) and 2.45x (napi) faster than the JSON transport.
- Profiled remaining cost (examples/profile_stages.rs): heart-rate ACF in
  wifi-densepose-vitals dominates; left for a separate, output-reviewed
  change. Rejected WASM +simd128 and opt-level=s after three measured runs
  (bench/wasm-variants.mjs).

Device e2e (ADR-373 amendment):
- test/e2e/devices.e2e.mjs + `npm run test:e2e:devices` + new workflow
  ruview-device-e2e.yml: pty-emulated MR60BHA2 and RPLIDAR through real
  pyserial, a UDP ESP32 node, and the real iPhone relay, via the CLI.
- mmWave auto-detect keeps probe frames and reads only the remainder;
  iPhone LiDAR gains max_frames and returns immediately when refused.
  Suite wall time 19.6 s -> 5.6 s (MEASURED, dev container).

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_015H55jmprfz4oHWXSFbX5rf
2026-10-01 14:45:55 +00:00
Claudeandclaude-flow 6112a185df feat(npm): host device + SSH remote access — ESP32, 60/24 GHz mmWave, LiDAR (ADR-373/374)
@ruvnet/ruview 0.7.0 gives CLI, MCP, and SDK access to RuView sensing
hardware on the machine it is attached to, and to other machines over SSH.

Devices (ADR-373, src/devices/):
- ruview_devices_scan: pyserial port list classified by USB VID:PID
  (Espressif native USB, CP210x, CH34x, FTDI) into ESP32 / mmWave / RPLIDAR
  roles with the command that confirms each.
- ruview_esp32_capture: receive-only UDP listener parsing ADR-018 CSI
  frames, vitals, and the other 0xC511000x packets per node: rate,
  sequence loss, RSSI, subcarriers, frequency, device vitals.
- ruview_mmwave_read: MR60BHA2 (60 GHz) and LD2410 (24 GHz) parsers ported
  byte-for-byte from firmware mmwave_sensor.c; baud/frame auto-detect.
- ruview_lidar_read: RPLIDAR SCAN over serial (start/stop, DTR best-effort)
  and the iPhone LiDAR relay as a WebSocket client (token only via
  RUVIEW_LIDAR_TOKEN; statistics only, never raw depth).
- Dependency-free serial via a fixed pyserial pump (argv-only, bounded,
  hex output immune to secret redaction). Schema-level port pattern
  validation added to the policy validator.
- All device tools require the new `device-access` MCP grant.

Remote hosts (ADR-374, src/remote.js):
- ~/.config/ruview/hosts.json (0600), CLI-only `hosts add`; strict
  name/target validation against ssh option and shell injection.
- `ssh -o BatchMode=yes -o StrictHostKeyChecking=yes -- <target>` running a
  version-pinned `ruview call <tool> --read-only`, every token single-quoted;
  read-only enforced locally and remotely; no mutations, credentialed reads,
  or host hopping. MCP needs `remote-host` plus the remote tool's grant.

Also: device-access skill, doctor `devices`/`remote` groups, SDK
`devices.*`/`hosts.*` with types, generic `ruview call` verb, guidance
capability, 320 KiB package budget, CI smoke expects 22 tools.

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_015H55jmprfz4oHWXSFbX5rf
2026-09-30 17:01:15 +00:00
Claudeandclaude-flow 888222277e feat(npm): RuView operator surface — SDK, cross-platform flashing, training, doctor (ADR-369..372)
@ruvnet/ruview 0.6.0 turns the harness into a complete npm operator
surface: one policy-checked tool registry behind the CLI, the MCP server,
and a new typed SDK.

- SDK (ADR-369): `@ruvnet/ruview/sdk` createRuView() with doctor,
  firmware.{ports,plan,flash,monitor}, training.{plan,run,gate,calibrate},
  kernel.{selfTest,load}, guidance/claimCheck/verify/spaces; strict mode
  throws RuViewError; TypeScript declarations in src/sdk.d.ts.
- Firmware (ADR-370): cross-platform `python -m esptool` flashing replaces
  the Windows-only stub. SHA256SUMS verification (mismatch always fatal),
  bundle containment, variant/partition checks, argv-only command, chip_id
  preflight that refuses a mismatched chip, boot-log capture;
  hardwareValidated only when CSI callbacks appear without a panic.
  Read-only ruview_firmware_plan / ruview_firmware_ports.
- Training (ADR-371): ruview_train / ruview_train_plan run pose-smoke,
  pose (MM-Fi), or room training from the trusted checkout with in-repo
  paths only, logs/checkpoints under v2/target/ruview-train, and libtorch
  remedies. ruview_train_gate enforces mean-pose baseline, leakage-free
  splits, subject/time disjointness, and evidence tagging.
- Doctor (ADR-372): structured pass/warn/fail/skip checks with remedies
  across runtime, integrity, hosts, repo, Rust, Python/esptool, firmware
  bundles, serial, sensing server, and the compute kernel; --json/--group;
  hardware probe and network checks are CLI/SDK-only.
- Plan tools cannot take `confirm`; acting tools keep grant+confirm over
  MCP. Policy files and TOOL_POLICY are drift-tested.
- Unpacked budget 160 KiB → 224 KiB in both npm workflows; CI tarball smoke
  now checks the SDK and the training gate.

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_015H55jmprfz4oHWXSFbX5rf
2026-09-30 16:47:30 +00:00
Claudeandclaude-flow 1938f0251f feat(npm): @ruvnet/ruview-kernel — WASM + napi-rs compute kernel with metaharness e2e (ADR-368)
Adds a portable RuView compute kernel for JavaScript, built from the Rust
source of truth and wired into the @ruvnet/ruview metaharness.

Rust
- v2/crates/ruview-kernel: bounded single-function JSON ABI
  (call(op, json) -> envelope) over the ADR-021 wifi-densepose-vitals
  pipeline: streaming sessions, batch analyze, deterministic SplitMix64
  synthetic CSI (SYNTHETIC), bounded session registry, strict input
  validation (deny_unknown_fields, width/finite/magnitude checks, size and
  count limits). wasm32-only zero-import C ABI (rvk_alloc/free/call).
- v2/crates/ruview-kernel-napi: napi-rs transport exposing only
  abiVersion()/call(); standalone workspace (excluded from v2) so napi never
  enters `cargo test --workspace`; panic=unwind + catch_unwind.

npm (harness/ruview-kernel, zero runtime deps)
- loadKernel({backend: wasm|napi|auto}); WASM default (no host authority),
  napi never silently falls back, auto reports fallbackReason.
- SHA256SUMS integrity verification; mismatch is always fatal; loader
  rejects any WASM module that imports host functions.
- Browser-safe ./wasm entry; CLI doctor/info/selftest/parity/synth/
  analyze/bench; build.mjs builds artifacts from source (no shell).
- Unit, security, and e2e tests (CLI round trip, cross-backend parity).

Metaharness
- New read-only tool ruview_kernel_selftest / `ruview kernel`: imports the
  fixed optional package, fails closed with kernel_not_installed; policy,
  claims, MCP policy, guidance capability, and manifest updated.

CI: .github/workflows/ruview-kernel-npm.yml builds both transports on
Linux/macOS/Windows, requires napi in tests, runs parity, verifies
artifacts, packs, and runs the cross-package e2e. No publish step.

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_015H55jmprfz4oHWXSFbX5rf
2026-09-30 16:23:29 +00:00
rUv b90b592d4f fix(vitals): reject respiratory peaks and cap filtered confidence (#2061) 2026-09-30 10:20:41 -04:00
rUv e1fbbede61 fix(sensing): exclude stale nodes from room occupancy (#2060) 2026-09-30 09:25:20 -04:00
rUv aac4155584 feat(bfi): add validated capture and bounded research tooling (#2055)
* feat(bfi): add validated capture and bounded research tooling

* fix(deps): patch Wasmtime for RUSTSEC-2026-0316
2026-09-29 23:53:48 -04:00
Dragan Spiridonov 5ef001b4fe Merge pull request #1904 from ruvnet/codex/firmware-delivery-20260911
fix(firmware): restore native C6 onboarding delivery
2026-09-27 11:43:55 +02:00
rUv 6524712900 Merge pull request #1890: storage partition for 8MB, compile node_log out on 4MB
Verified on a physical ESP32-C6 (4 MB, esp32c6 defaults, ESP-IDF v5.4):
- App image: 1,085,424 B on main, 1,042,368 B with this change (43,056 B smaller).
- The 'Failed to find FATFS partition storage' error seen on every 4 MB C6 boot is gone.
- Wi-Fi join, CSI init and CSI UDP delivery to the host were confirmed.
CI: 35/35 green.
2026-09-26 12:36:57 -05:00
rUv dd02efe2fe Merge pull request #1903 from ruvnet/codex/calibration-process-identity
fix(calibration): bind server receipts to room identity
2026-09-16 16:30:28 +00:00
ruv 7d15e9fb4a fix(calibration): register calibrated-presence-evidence schema via field_bridge
The new schema constant was defined and asserted as a raw string literal
directly in main.rs, which the semconv registry test scans and requires
every "ruview.*" literal there to be a registered attribute/event. This
isn't a tracing event, though -- it's a wire-schema tag structurally
identical to CALIBRATION_MODEL_RECEIPT_SCHEMA, which already lives in
field_bridge.rs specifically to stay out of that scan. Move the new
constant there and reference it by identifier from main.rs, matching the
existing precedent instead of hand-editing the generated semconv registry.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-09-16 11:53:26 -04:00
ruv 69106970dd Merge branch 'main-ref-tmp' into codex-calib-work 2026-09-16 11:37:53 -04:00
ruv 686e4bad00 Merge branch 'fix-1936-work' into codex-calib-work 2026-09-16 11:37:42 -04:00
ruv b713a23cac fix(calibration): preserve expired status instead of collapsing to none
calibration_status unconditionally overwrote status with "none" whenever
active was false, which also erased the legitimate "expired" status for
a calibration that ran long ago and has since expired -- losing the
distinction between "never calibrated" and "was calibrated, now expired".

Fixes the pre-existing failure in
calibration_expiry_tests::long_running_expiry_is_not_reported_active_for_runtime_or_bootstrap,
called out as a known, unrelated gap in PR #1936.

Co-Authored-By: claude-flow <ruv@ruv.net>
2026-09-16 11:37:31 -04:00
Dragan Spiridonov 0c1efda425 chore(firmware): pin exact IDF 5.4 C6 artifact 2026-09-16 09:58:16 +02:00
Dragan Spiridonov b1be91478c fix(firmware): restore native C6 onboarding delivery 2026-09-16 09:58:16 +02:00
Joe 6eb79cad26 chore: empty commit to re-trigger stale CI (row 35 sweep, main@45fc41ac) 2026-09-15 22:28:35 -04:00
rUv 45fc41ac4c Merge pull request #1944 from ruvnet/codex/realtek-csi-upstream-main
Add bounded RTL8721Dx RAC1 CSI ingest to current main
2026-09-15 19:14:52 +00:00
rUv 901cf16eb3 Merge pull request #1945 from ruvnet/codex/adr-348-collision-fix
Repair forecast ADR collision and RustSec lockfile gate
2026-09-15 18:56:09 +00:00
ruv af2d6981a3 ci: bound S3 8 MB image to 1200 KiB 2026-09-15 14:41:08 -04:00
ruv a04fbc4694 Resolve ADR 348 collision in upstream forecast CI 2026-09-15 14:41:08 -04:00
ruv dbaa3c7f52 ci: bound S3 8 MB image to 1200 KiB 2026-09-15 14:16:05 -04:00
ruv 6cf0356554 Update rustls lockfile for RUSTSEC-2026-0285 2026-09-15 14:07:11 -04:00
ruv b2ad3a65dc Resolve ADR 348 collision in upstream forecast CI 2026-09-15 14:06:31 -04:00
ruv 815384dcc4 Update rustls lockfile for RUSTSEC-2026-0285 2026-09-15 14:02:38 -04:00
ruv 8cd8742d2b Add bounded RTL8721Dx RAC1 CSI ingest to current RuView main 2026-09-15 13:57:01 -04:00
Dragan SpiridonovandRuflo & AQE dbf4788628 fix(calibration): bind the grid the radio keeps emitting, not the sparse minority
`accept_grid` locks each node onto the densest grid it has seen and rejects
sparser frames from the feature path -- on an ESP32-C6 the ~16% HT 64-bin
minority that arrives alongside HE-SU 256-bin. `select_calibration_grid`
ordered candidates by `max_gap_s` first, so it preferred whichever grid
looked temporally smoothest. A sparse trickle at a metronomic cadence has a
smaller worst-case gap than a busy stream with one scheduling hiccup, so
selection kept choosing the grid admission is designed to discard.

Measured on four ESP32-C6 nodes: a capture bound 64sc on node 11, every node
then locked onto 256sc, the bound grid went stale with no frame for five
minutes, frame_count froze at 10,997 and `calibration_grid_is_fresh` failed.
The capture could never finalize -- `collecting` for 858 s with both minimums
long since met, no error surfaced, and a flickering UI as the client retried.
The same rig bound 256sc on a different capture and finalized with five
baseline eigenvalues.

Order by subcarrier count first so selection agrees with admission, then fall
back to the previous gap/rate/ppdu ordering to choose among equally dense
candidates.

Co-Authored-By: Ruflo & AQE
Claude-Session: https://claude.ai/code/session_01Qh7eir6C2U5CSwD7GR43Zv
2026-09-15 14:48:08 +02:00
Dragan SpiridonovandRuflo & AQE cdad032296 fix(calibration): only the grid-bound radio may write the single-link baseline
A room binding may name several radios, and every one of them was admitted
into the field model. That model is single-link: one baseline, one set of
amplitude offsets. Averaging two radios' offsets into it flattens the
eigenstructure and leaves only a scalar energy threshold behind.

Measured on three ESP32-C6 nodes:

  4-node binding  baseline_eigenvalue_count 0, residual threshold 18.90,
                  empty room reported an occupant in 1162/1162 frames
  1-node binding  baseline_eigenvalue_count 5, residual threshold 3.65,
                  empty room absent 604/604, occupied present 616/616

`person_count_at` already refused to score a non-bound radio, calling it
"deterministic false occupancy from hardware-specific amplitude offsets" --
but only when exactly one node was bound. With more it fell through to the
shared mixed history, so the guard never fired in the case it described.

Restrict the baseline write, not the admission. Every bound radio is still
admitted and recorded as a contributor: `calibration_source_nodes_missing`
refuses to finalize until each one proves it is live on the frozen grid, so
gating them out of the feed path entirely deadlocks the capture -- measured
on hardware as 43,120 frames over 51 minutes stuck in `collecting` with
missing_source_node_ids=[12,13,14]. Only `maybe_feed_calibration` is
restricted to the grid-bound radio, and scoring follows that same radio
whatever the size of the room's node set.

This is the narrowing `bootstrap_baseline::store` already applies when it
persists `vec![binding.source_node_id]` as the frozen model source.

Co-Authored-By: Ruflo & AQE
Claude-Session: https://claude.ai/code/session_01Qh7eir6C2U5CSwD7GR43Zv
2026-09-15 14:09:33 +02:00
Dragan SpiridonovandRuflo & AQE 0decc5c3fa test(calibration): cover calibrated presence evidence identity and refusals
Assert the exact wire schema and every identity field the Mac app matches
against its model receipt, plus that evidence agrees with the server's own
person count. Pair the refusals (no receipt, bootstrap-only authority,
expired model) with the positive case so an absent value can never be read as
proof when the path is simply broken.

Co-Authored-By: Ruflo & AQE
Claude-Session: https://claude.ai/code/session_01Qh7eir6C2U5CSwD7GR43Zv
2026-09-14 18:54:15 +02:00
Dragan SpiridonovandRuflo & AQE d4a3c7f17f feat(calibration): emit calibrated presence evidence v2 on the sensing stream
The Mac app gates its held-out empty-room check on
`calibrated_presence_evidence` carrying
`ruview.calibration.calibrated-presence-evidence.v2`. No shipped server has
ever emitted that key, so the check fails on its first iteration with "No
fresh calibrated presence result matched this room's current model receipt",
`bootstrap/promote` is never called, and `bootstrap_baseline::store` -- the
only path that persists a calibration -- never runs. A completed 62-minute
field model is therefore lost on the next restart.

Emit the evidence bound to the active model receipt. `None` unless an
explicit calibration is fresh and the field model scores the observation
without a heuristic fallback, so a present value is always calibrated.

Extract the field-model scoring history into `scoring_history()` and reuse it
for both `person_count_at` and the evidence, so the published evidence cannot
disagree with the count derived from the same model.

Co-Authored-By: Ruflo & AQE
Claude-Session: https://claude.ai/code/session_01Qh7eir6C2U5CSwD7GR43Zv
2026-09-14 18:49:50 +02:00
Dragan Spiridonov 2c0abcc802 fix(calibration): separate receipt schema from telemetry literals 2026-09-11 20:29:42 +02:00
Dragan Spiridonov f1dbaac452 Merge remote-tracking branch 'origin/main' into codex/calibration-process-identity 2026-09-11 20:27:02 +02:00
rUv 33a9e90896 Merge pull request #1902 from ruvnet/feat/firmware-usb-onboarding-c6
firmware: USB onboarding on main + C6 console RX fix (0.8.12)
2026-09-11 13:32:02 +00:00
Dragan Spiridonov bbbb14169d fix(calibration): bind server receipts to room identity 2026-09-11 15:21:39 +02:00
ruv 4fbe58f736 fix(firmware): keep onboarding digest compatible with ESP-IDF 6 2026-09-11 09:19:33 -04:00
ruvandClaude Fable 5.1 a64e9e98fb fix(firmware): keep the C6 console RX alive and ship USB onboarding (0.8.12)
- mmwave_sensor: the default probe pins (UART1 TX=17/RX=18) put UART1 TX on
  GPIO17, which is U0RXD on the ESP32-C6. Console input was dead on every C6
  node even when no sensor was found, because uart_driver_delete() leaves the
  GPIO matrix routing in place. C6 now defaults to GPIO4/5 and any pin pair
  overlapping the console UART is refused.
- serial_onboarding_protocol: _POSIX_C_SOURCE for strtok_r under -std=c11
  -pedantic (host tests on glibc).
- main/CMakeLists: explicit led_strip REQUIRES so builds work with the
  component manager disabled (vendored components, offline builds).
- release_bins/c6-onboarding: 0.8.12 C6 images built with ESP-IDF v5.4.

Hardware-verified on an ESP32-C6 over cu.usbserial-3120 from the RuView Mac
app: flash -> RUVIEW_HELLO_V1 receipt -> RUVIEW_CONFIG_V1 -> node 3 live at
-34 dBm on the sensing server.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Rqz356q3Qz6sscTrpy5Jrv
2026-09-11 08:58:46 -04:00
ruv 130afabae8 feat(firmware): add secure macOS USB onboarding
(cherry picked from commit 3373e79f38afef3a1430feeb78ae034e78777c66)
2026-09-10 19:25:13 -04:00
Joeandclaude-flow 36e4df95b7 feat(firmware): give node_log an 8MB home, compile it out on 4MB
node_log currently only has a FAT volume on the 16MB table
(partitions_16mb.csv), so it silently does nothing on the two chips
most of the fleet's OTA/CI surface actually exercises.

partitions_display.csv (8MB) ends at 0x600000 with 2.00 MB of flash
never allocated. Add a storage/fat partition filling exactly that
tail (0x600000 + 0x200000 = 0x800000, the full 8 MB with nothing left
over), so node_log has somewhere to mount on the s3/8mb profile.

partitions_4mb.csv has no equivalent room -- two 1.81 MB app slots
leave only 256 KB free, not enough for a FAT volume of any useful
size. Guard node_log.c's whole implementation on
CONFIG_ESPTOOLPY_FLASHSIZE_4MB and provide no-op stubs for every
entry point, so main.c's call sites need no #ifdef of their own.
Deliberately did not touch main/CMakeLists.txt: fatfs/wear_levelling
stay in REQUIRES, and with nothing left calling into them on 4MB,
--gc-sections drops both archives the same way it already drops
libespcoredump.a when coredump-to-flash is unused (measured at 0
bytes contribution).

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_01PVWMiHQifoYXL7uL3bphrZ
2026-09-10 13:21:59 -04:00
Faizanandgyrwb6vjbj-ctrl b370d4fcd4 fix: enable macOS RSSI sensing via CoreWLAN (#1845)
* fix: enable macOS RSSI sensing via CoreWLAN

* fix: abstain when macOS WiFi identity is unavailable

---------

Co-authored-by: gyrwb6vjbj-ctrl <253160963+gyrwb6vjbj-ctrl@users.noreply.github.com>
2026-09-10 00:15:02 +00:00
500cfe4787 feat(firmware): persistent on-node log and coredump-to-flash (#1878)
* feat(firmware): add the 16MB flash profile the C6 fleet runs on

The 4MB and display layouts have 1.875MB and 2MB OTA slots, sized for 4MB
and 8MB parts. A 16MB C6 (DevKitC-1 N16) has no layout in the tree at all,
so building for one meant hand-editing sdkconfig -- and a smaller table
written over a deployed 16MB node relocates NVS and destroys its
provisioning, which is why this belongs in the repo rather than in a
workstation's untracked sdkconfig.

Two 4MB OTA slots, plus a 64K coredump partition and an 8000K FAT volume
for on-node storage. Rollback is enabled so an OTA'd image that cannot
confirm itself is reverted by the bootloader.

Layered, not inherited: sdkconfig.defaults.esp32c6 keeps its 4MB layout so
the common 4MB devkit is never silently repartitioned.

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_01PVWMiHQifoYXL7uL3bphrZ

* feat(firmware): persistent on-node log and coredump-to-flash

A remote fault currently leaves no post-mortem. main.c decodes the reset
reason at boot and logs it to the CONSOLE -- which is exactly what a power
cycle destroys, and a power cycle is how a wedged node gets recovered.

The flash was already provisioned for this and the features were simply off:
partitions_16mb.csv has reserved a 64K `coredump` partition and an 8000K
`storage` FAT volume since the 16MB layout was written, neither had ever been
mounted or written, and the build defaulted to CONFIG_ESP_COREDUMP_ENABLE_TO_NONE.

- node_log.c/h: a 128 B-record ring on the storage volume. Boot records
  (reset reason + the previous session's final uptime + a health snapshot),
  periodic health samples, and event records. Ordering is boot_id + uptime_s
  because these boards have no RTC and nothing can be timestamped absolutely.
  Counters are stored RAW and differenced at read time, so a cadence change
  never invalidates old records. Retention, not wear, binds: 128 B every
  300 s is ~208 days in 8 MB, and the write rate is bounded explicitly.
- ap_rssi_dbm, the association link from esp_wifi_sta_get_ap_info. The
  adaptive controller's rssi_median_dbm is a CSI statistic, not the uplink,
  so a node with a bad uplink currently shows no signal for it anywhere.
- CONFIG_ESP_COREDUMP_ENABLE_TO_FLASH, complementary rather than a
  substitute: a coredump catches a CRASH, the boot record catches a HANG
  followed by a watchdog reset.

node_log_init failing is non-fatal by design -- the module disables itself
and every entry point becomes a no-op. A logging volume that will not mount
must never take the fleet down with it.

csi_collector gains two gate accessors. This build selects its frame gate at
compile time, so both return CSI_GATE_NOT_CONFIGURABLE rather than a
plausible number: the log keeps a field of fixed width and meaning either
way, and an explicit "not applicable" is stored instead of something a
reader would later reason from as a real setting.

Built with ESP-IDF v5.4 for esp32c6 using
defaults + .esp32c6 + .16mb. Verified from the resulting artifacts: 16MB
table with coredump and storage present, COREDUMP_ENABLE_TO_FLASH set,
rollback enabled, image 0x10abd0 bytes in a 4M slot. NOT flashed to hardware
from this branch -- the equivalent code was exercised on a node-3 pilot, but
that is not evidence for this build.

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_01PVWMiHQifoYXL7uL3bphrZ

* docs(adr-362): reading the on-node log is on-demand, never streamed

The log engine lands with no consumer, so the access pattern was open. Records
the decision to pull a node's log only when a human asks -- after a hang or a
crash, or while troubleshooting that node -- and to keep nothing server-side.

The reasoning that decided it: the ring already survives a power cycle on the
node (~208 days of retention), so a server copy duplicates a durability
property that is already met; and airtime is this fleet's binding constraint,
measured at 91% consumed by two slow nodes, so a periodic drain across nine
boards would spend exactly the resource whose exhaustion you would open the
log to investigate. An instrument must not perturb what it measures.

Leaves the decode-where question open on purpose, with the argument for a raw
proxy stated: decoding server-side defines the record layout in both C and
Rust, and the pilot already paid once for a host/firmware disagreement about a
field's meaning.

Co-Authored-By: claude-flow <ruv@ruv.net>
Claude-Session: https://claude.ai/code/session_01PVWMiHQifoYXL7uL3bphrZ

---------

Co-authored-by: Joe <joe@GitDev.hedgehog>
Co-authored-by: claude-flow <ruv@ruv.net>
2026-09-09 23:50:34 +00:00
f7c2c6587e feat(firmware): add the 16MB flash profile the C6 fleet runs on (#1877)
The 4MB and display layouts have 1.875MB and 2MB OTA slots, sized for 4MB
and 8MB parts. A 16MB C6 (DevKitC-1 N16) has no layout in the tree at all,
so building for one meant hand-editing sdkconfig -- and a smaller table
written over a deployed 16MB node relocates NVS and destroys its
provisioning, which is why this belongs in the repo rather than in a
workstation's untracked sdkconfig.

Two 4MB OTA slots, plus a 64K coredump partition and an 8000K FAT volume
for on-node storage. Rollback is enabled so an OTA'd image that cannot
confirm itself is reverted by the bootloader.

Layered, not inherited: sdkconfig.defaults.esp32c6 keeps its 4MB layout so
the common 4MB devkit is never silently repartitioned.


Claude-Session: https://claude.ai/code/session_01PVWMiHQifoYXL7uL3bphrZ

Co-authored-by: Joe <joe@GitDev.hedgehog>
Co-authored-by: claude-flow <ruv@ruv.net>
2026-09-09 23:31:03 +00:00
30391bfbd2 docs(adr-361): single-transmitter pairing -- hypothesis confirmed, fix not adopted (#1876)
A-B-A capture from the overnight run, ten 30 s windows per arm: pairing
fraction rises 0.5668 -> 0.7287 (Welch t=13.26, p<0.0001) when the fleet is
filtered to one transmitter, but paired observations/s falls 44% because
tx/s falls 56%. Fraction was the wrong figure of merit; not adopted.

Firmware changes (persistent on-node log, live filter_mac, and the node-3
pilot fixes) that produced this capture are held back pending a port against
main's restructured firmware/main/ tree (config_api.c and ota_update.c are
gone there).


Claude-Session: https://claude.ai/code/session_01PVWMiHQifoYXL7uL3bphrZ

Co-authored-by: Joe <joe@GitDev.hedgehog>
Co-authored-by: claude-flow <ruv@ruv.net>
2026-09-09 23:31:00 +00:00
rUv 00a42d7cda fix(sensing-server): don't let edge-vitals sentinel RSSI poison mean_rssi (#1882)
The ESP32 edge-vitals packet (0xC511_0002) carries a raw i8 RSSI byte with
no "valid" flag. When the edge pipeline hasn't sampled a real reading yet,
it has been observed sending near-zero sentinel values (-1, -2 dBm) that
are physically implausible for a WiFi link. These were pushed straight
into the node's rssi_history and the room's fused mean_rssi, producing a
one-tick reading of ~-2 dBm on node 3 while its own CSI path simultaneously
showed a real ~-50 dBm signal.

Add an is_plausible_rssi() guard (mirroring the existing
update_csi_fps_ema() plausibility-filter pattern) so an implausible RSSI
falls back to the node's last known-good reading instead of overwriting it.
2026-09-09 20:10:06 +00:00
rUv bd6f38e2df fix(engine): wire node positions into StreamingEngine's governed fuser (#1880)
StreamingEngine owns its own MultistaticFuser instance -- a different one
from wifi-densepose-sensing-server's AppState, which was fixed for this
same class of bug in #1860. Nothing ever called set_node_positions /
set_node_positions_by_id on this instance, so every FusedSensingFrame
produced by the governed path (fuse_scored_calibrated, which feeds
provenance and the witness) carried every node's position as the origin
regardless of deployment configuration.

Adds StreamingEngine::set_node_positions(HashMap<u8, [f32;3]>), delegating
to the fuser's existing set_node_positions_by_id. Also fixes a related
hazard in set_multistatic_config: it rebuilt the fuser wholesale
(MultistaticFuser::with_config(cfg)), which silently discarded any
positions configured before it. It now preserves them across the rebuild,
so callers don't need to remember a specific ordering between the two
setters.

Two regression tests per the issue's own acceptance criteria: one calling
fuse_scored_calibrated directly and asserting each fused contribution
carries its own node's configured position, and one asserting positions
survive a later set_multistatic_config call.

Fixes #1870.
2026-09-09 17:28:06 +00:00
rUv 203bd66daa Merge pull request #1879 from ruvnet/fix/hf-csi-embedding-metrics
fix(training): export the real contrastive lossHistory/duration, not a no-op baseline pass
2026-09-09 16:27:40 +00:00
ruv 2c3a924231 fix(training): export the real contrastive lossHistory/duration, not a no-op baseline pass
scripts/train-ruvllm.js ran two separate computations during contrastive
pretraining: ContrastiveTrainer.train() (a baseline pass that computes loss
but never updates the encoder) and a hand-written loop that actually applies
gradient updates to encoder.w2. The exported training-metrics.json took
lossHistory and contrastive.durationMs from the first (no-op) pass instead
of the second (real) one, so the published curve was flat/meaningless
(~0.1352 for all 20 epochs) and the duration was ~30x too short.

This did not affect finalLoss/improvement or the held-out temporal-triplet
accuracy metric published for this model on HuggingFace
(ruvnet/wifi-densepose-pretrained) -- those already came from the real loop.
It only affected the diagnostic curve/duration anyone auditing convergence
from the published JSON would see.

Also: config.json's training.steps field (epochs * triplets, technically
consistent but reads like an optimizer-step count) is now reported as
separate epochs/tripletUpdatesPerEpoch fields, and the exported version
bumped 1.0.0 -> 2.0.1 to match the "v2" naming already used elsewhere for
this file set.

Verified by re-running against the real capture
(data/recordings/overnight-1775217646.csi.jsonl): lossHistory now shows a
genuine decreasing curve (0.1037 -> 0.0654 across 20 epochs, matching the
already-published finalLoss exactly), durationMs reports the real ~40s
gradient-loop time instead of ~12s.

Model card updated with the same disclosure (docs/huggingface/MODEL_CARD.md)
plus a real, previously-undisclosed EWC forgetting-rate figure (32.97%,
present in training-metrics.json but not surfaced in the human-readable
card) and a corrected repro-script pointer.
2026-09-09 11:26:17 -04:00
507 changed files with 62276 additions and 2308 deletions

No files matched your search

+5
View File
@@ -10,6 +10,11 @@
"name": "ruview",
"source": "./plugins/ruview",
"description": "End-to-end RuView toolkit: getting started, ESP32 hardware setup, configuration, sensing applications (presence / vitals / pose / sleep / MAT), camera-free + camera-supervised model training, advanced multistatic sensing, CLI / API / WASM, mmWave radar, and witness verification"
},
{
"name": "ruview-live",
"source": "./harness/ruview/mod",
"description": "Claude Code mod: /ruview opens a live pane with your CSI nodes (ESP32, Realtek) and ESPHome radar (presence, distance, device-reported vitals), refreshed on a timer, plus a one-line status. Read-only; ships inside @ruvnet/ruview. Mods are early access (CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1)."
}
]
}
+4
View File
@@ -135,6 +135,10 @@ async function runProcess(command, args, {
resolve(result);
});
});
// A child that never reads stdin (e.g. `git check-ignore`) may exit and
// close the pipe before this write flushes; the resulting EPIPE is not a
// failure of the command, whose real outcome the 'close' handler reports.
child.stdin.on('error', () => {});
child.stdin.end(input);
});
}
+43
View File
@@ -0,0 +1,43 @@
name: BFI Research Tools
on:
pull_request:
paths:
- 'tools/bfi/**'
- 'tools/bfi_learning/**'
- '.github/workflows/bfi-research.yml'
push:
branches: [main]
paths:
- 'tools/bfi/**'
- 'tools/bfi_learning/**'
- '.github/workflows/bfi-research.yml'
workflow_dispatch:
permissions:
contents: read
concurrency:
group: bfi-research-${{ github.ref }}
cancel-in-progress: true
jobs:
offline-tests:
name: BFI offline tests (Linux, CPU)
runs-on: ubuntu-latest
timeout-minutes: 15
steps:
- uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262
with:
persist-credentials: false
- uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
with:
python-version: '3.13'
- name: Install numeric test dependencies
run: python -m pip install numpy==2.4.2 scipy==1.17.1
- name: Install CPU PyTorch
run: python -m pip install torch==2.11.0 --index-url https://download.pytorch.org/whl/cpu
- name: Test capture parsing and bounded runners
run: python -B -m unittest discover -s tools/bfi -v
- name: Test offline learning and data admission
run: python -B -m unittest discover -s tools/bfi_learning -v
+14 -7
View File
@@ -204,7 +204,10 @@ jobs:
node-version: '22'
- name: Run UI unit tests
run: node --test ui/sw.test.mjs ui/services/ws-ticket.test.mjs ui/services/websocket.service.test.mjs v2/crates/wifi-densepose-desktop/ui/build-config.test.mjs
run: node --test ui/sw.test.mjs ui/services/ws-ticket.test.mjs ui/services/websocket.service.test.mjs ui/services/sensing.service.test.mjs ui/components/DashboardTab.test.mjs v2/crates/wifi-densepose-desktop/ui/build-config.test.mjs
- name: Run virtual-view adapter boundary tests
run: node --test integrations/virtual-view/adapter.test.mjs
# Unit and Integration Tests
# Python pytest matrix — runs against the archived v1 Python tree.
@@ -501,7 +504,7 @@ jobs:
notify:
name: Notify
runs-on: ubuntu-latest
needs: [code-quality, test, rust-tests, performance-test, docker-build, docs]
needs: [code-quality, test, rust-tests, ui-tests, performance-test, docker-build, docs]
if: always()
permissions:
contents: write # required by softprops/action-gh-release
@@ -528,18 +531,22 @@ jobs:
text: '❌ CI pipeline failed for ${{ github.ref }}'
- name: Create GitHub Release
if: github.ref == 'refs/heads/main' && needs.docker-build.result == 'success'
if: github.ref == 'refs/heads/main' && needs.rust-tests.result == 'success' && needs.ui-tests.result == 'success' && needs.docker-build.result == 'success'
uses: softprops/action-gh-release@3bb12739c298aeb8a4eeaf626c5b8d85266b0e65
with:
tag_name: v${{ github.run_number }}
name: Release v${{ github.run_number }}
name: CI snapshot v${{ github.run_number }}
# CI snapshots must not replace a curated product release as Latest.
make_latest: false
body: |
Automated release from CI pipeline
Source snapshot validated by the CI pipeline. This entry does not contain firmware bundles.
**Changes:**
${{ github.event.head_commit.message }}
**Docker Image:**
`${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}:${{ github.sha }}`
**Source commit:** `${{ github.sha }}`
Container images publish separately through the [sensing-server release workflow](https://github.com/ruvnet/RuView/actions/workflows/sensing-server-docker.yml).
Firmware downloads are attached to the dedicated `vX.Y.Z-esp32` releases.
draft: false
prerelease: false
+57 -6
View File
@@ -16,6 +16,17 @@ on:
- '.github/workflows/firmware-ci.yml'
jobs:
provisioning-tests:
name: Provisioning tooling tests
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262
- uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
with:
python-version: '3.11'
- name: Verify provisioning and secret handling without hardware
run: python -m unittest discover -s firmware/esp32-csi-node/tests -p 'test_provision*.py' -v
version-guard:
name: Verify version.txt matches release tag
runs-on: ubuntu-latest
@@ -53,7 +64,9 @@ jobs:
sdkconfig: sdkconfig.defaults
partition_table_name: partitions_display.csv
size_warn_kb: 1100
size_limit_kb: 1152
# Current 8 MB image is 1163 KiB after USB onboarding. The OTA app
# partition is 2048 KiB; retain a 1200 KiB hard gate and warning.
size_limit_kb: 1200
artifact_app: esp32-csi-node.bin
artifact_pt: partition-table.bin
- variant: 4mb
@@ -92,14 +105,11 @@ jobs:
idf.py set-target ${{ matrix.target }}
idf.py build
- name: Build and run host-side ADR-110 unit tests
- name: Build and run host-side firmware unit tests
if: matrix.variant == 'c6-4mb'
working-directory: firmware/esp32-csi-node/test
run: |
make test_adr110
./test_adr110
make test_c6_antenna
./test_c6_antenna
make host_tests
- name: Verify binary size budget
working-directory: firmware/esp32-csi-node
@@ -120,6 +130,10 @@ jobs:
fi
echo "Binary size OK: $SIZE <= $MAX"
- name: Verify built application version before packaging
working-directory: firmware/esp32-csi-node
run: python test/test_delivery_contract.py --built-app build/esp32-csi-node.bin
- name: Verify flash image integrity
working-directory: firmware/esp32-csi-node
run: |
@@ -205,3 +219,40 @@ jobs:
idf.py build
test "$(stat -c%s build/esp32-csi-node.bin)" -le $((1152 * 1024))
strings build/esp32-csi-node.bin | grep -F "external U.FL"
publish-release-assets:
name: Publish firmware release assets
if: github.ref_type == 'tag'
needs: [version-guard, provisioning-tests, build, build-c6-idf6]
runs-on: ubuntu-latest
permissions:
contents: write
steps:
- name: Download the exact-run firmware artifacts
uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093
with:
pattern: esp32-csi-node-firmware-*
path: release-downloads
- name: Package variants without filename collisions
run: |
mkdir -p release-assets
for directory in release-downloads/esp32-csi-node-firmware-*; do
variant="${directory##*/}"
tar -C "$directory" -czf "release-assets/${variant}.tar.gz" .
done
(cd release-assets && sha256sum *.tar.gz > SHA256SUMS.txt)
- name: Publish the downloadable firmware bundles
uses: softprops/action-gh-release@3bb12739c298aeb8a4eeaf626c5b8d85266b0e65
with:
name: ESP32 firmware ${{ github.ref_name }}
prerelease: true
make_latest: false
body: |
Exact-run ESP32-S3 display 8 MB, ESP32-S3 headless 4 MB, and ESP32-C6 4 MB firmware bundles.
Verify each archive against `SHA256SUMS.txt` before flashing.
CI builds and host tests verify software behavior, not target hardware.
This prerelease has no new physical-board qualification or sensing-accuracy claim.
Select the matching board configuration using firmware/esp32-csi-node/README.md.
files: release-assets/*
+15 -5
View File
@@ -41,8 +41,12 @@ jobs:
build: false
publishable: true
# ADR-283/325: brain + local hosts + replay assets + guarded Spaces OAuth adapter;
# still runtime-dependency-free. 160 KiB is the reviewed hard ceiling.
unpacked_budget: 163840
# still runtime-dependency-free. ADR-369..374 add the SDK, firmware
# flashing, training runner/gate, doctor, device access, and SSH
# hosts. ADR-375 adds the MCP Apps console widget (~12 KB), the
# Streamable HTTP transport and the terminal UI (384 KiB); ADR-378 the
# ruview-live showcase views (~37 KB): 448 KiB reviewed ceiling.
unpacked_budget: 458752
- dir: harness/homecore
build: false
publishable: true
@@ -131,13 +135,19 @@ jobs:
npm i --no-fund --no-audit "$TGZ"
case "${{ matrix.package.dir }}" in
harness/ruview)
./node_modules/.bin/ruview --version
./node_modules/.bin/ruview doctor
./node_modules/.bin/ruview-harness --version
./node_modules/.bin/ruview-harness doctor
# the honesty gate must fail closed on empty input (ADR-263 F1)
if ./node_modules/.bin/ruview claim-check; then
if ./node_modules/.bin/ruview-harness claim-check; then
echo 'claim-check passed with no input — fail-open regression'; exit 1
fi
node --input-type=module -e "const m = await import('@ruvnet/ruview'); if (!m.TOOLS) process.exit(1);"
# ADR-369: the typed SDK resolves and the evidence gate refuses a baseline-free score
node --input-type=module -e "const { createRuView } = await import('@ruvnet/ruview/sdk'); const r = createRuView(); if (r.tools().length < 22) process.exit(1); const g = await r.training.gate({ model_score: 0.93 }); if (g.verdict !== 'FAIL') process.exit(1);"
./node_modules/.bin/ruview-harness doctor --json --group runtime,harness,firmware > doctor.json
if ./node_modules/.bin/ruview-harness train-gate --model-score 0.93 > /dev/null; then
echo 'train-gate passed without a baseline'; exit 1
fi
;;
harness/homecore)
./node_modules/.bin/homecore --version
+44
View File
@@ -0,0 +1,44 @@
name: README visual asset validation
on:
pull_request:
paths:
- 'README.md'
- 'assets/readme/**'
- 'docs/readme-visual-system.md'
- 'scripts/*readme*'
- 'scripts/tests/test_readme_visuals_xml.py'
- '.github/workflows/readme-visuals.yml'
push:
branches: [main]
paths:
- 'README.md'
- 'assets/readme/**'
- 'docs/readme-visual-system.md'
- 'scripts/*readme*'
- 'scripts/tests/test_readme_visuals_xml.py'
- '.github/workflows/readme-visuals.yml'
workflow_dispatch:
permissions:
contents: read
jobs:
verify:
name: Safe and reproducible README visuals
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262
- uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
with:
python-version: '3.11'
- name: Install visual tooling dependency
run: python -m pip install -r scripts/requirements-readme-visuals.txt
- name: Test XML input boundaries
run: python -m unittest discover -s scripts/tests -p test_readme_visuals_xml.py
- name: Verify reproducible and accessible assets
run: |
python scripts/render-readme-visuals.py
python scripts/verify-readme-visuals.py
git diff --exit-code -- assets/readme
test -z "$(git ls-files --others --exclude-standard -- assets/readme)"
+68
View File
@@ -0,0 +1,68 @@
# ADR-373/374 — device-access end-to-end through the real @ruvnet/ruview CLI.
# MR60BHA2 (60 GHz) and RPLIDAR are emulated on pseudo-terminals and read via
# real pyserial; an ESP32 node is emulated over UDP; the iPhone LiDAR path
# uses the repository's real relay. All results are SYNTHETIC: this proves
# the transports and parsers, not hardware behaviour.
name: ruview device e2e
on:
push:
branches: [main]
paths:
- 'harness/ruview/src/**'
- 'harness/ruview/bin/**'
- 'harness/ruview/test/e2e/**'
- 'integrations/iphone-lidar/web/**'
- '.github/workflows/ruview-device-e2e.yml'
pull_request:
paths:
- 'harness/ruview/src/**'
- 'harness/ruview/bin/**'
- 'harness/ruview/test/e2e/**'
- 'integrations/iphone-lidar/web/**'
- '.github/workflows/ruview-device-e2e.yml'
permissions:
contents: read
concurrency:
group: ruview-device-e2e-${{ github.ref }}
cancel-in-progress: true
jobs:
devices:
runs-on: ubuntu-latest
timeout-minutes: 15
steps:
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
with:
persist-credentials: false
- uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
with:
node-version: '22'
- uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
with:
python-version: '3.12'
- name: Install pyserial
run: python -m pip install pyserial==3.5
- name: Install the iPhone LiDAR relay
working-directory: integrations/iphone-lidar/web
run: npm ci --ignore-scripts
- name: Install harness
working-directory: harness/ruview
run: npm ci --ignore-scripts
# Missing prerequisites fail instead of skipping (RUVIEW_E2E_REQUIRE).
# The suite links each emulator pty to /dev/ttyRUVIEWE2E<n> with
# `sudo -n`, because the harness refuses /dev/pts/* port names.
- name: Device e2e (SYNTHETIC devices)
working-directory: harness/ruview
env:
RUVIEW_E2E_REQUIRE: '1'
run: npm run test:e2e:devices
+16 -2
View File
@@ -4,6 +4,12 @@ on:
pull_request:
paths:
- 'harness/ruview/**'
- 'scripts/csi-udp-relay.py'
- 'scripts/esp32_jsonl_to_rvcsi.py'
- 'scripts/overnight-empty-capture.py'
- 'scripts/record-csi-udp.py'
- 'scripts/collect-training-data.py'
- 'scripts/align-ground-truth.js'
- '.github/scripts/nightly-sota/**'
- '.github/workflows/nightly-sota-agent.yml'
- '.github/workflows/ruview-harness-flywheel.yml'
@@ -25,8 +31,12 @@ concurrency:
jobs:
verify:
name: Verify contributor harness
runs-on: ubuntu-latest
name: Verify contributor harness (${{ matrix.os }})
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
os: [ubuntu-latest, windows-latest]
defaults:
run:
working-directory: harness/ruview
@@ -39,9 +49,13 @@ jobs:
node-version: 22
cache: npm
cache-dependency-path: harness/ruview/package-lock.json
- uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
with:
python-version: '3.11'
- run: npm ci --ignore-scripts
- run: npm audit --omit=optional
- run: npm test
- run: npm run test:security
- run: npm run brain:verify
- run: npm run flywheel:plan
- run: npm run flywheel:verify
+132
View File
@@ -0,0 +1,132 @@
# ADR-368 — @ruvnet/ruview-kernel: build the WASM and napi-rs transports from
# source on every OS, run the Rust ABI tests, the package unit/security/e2e
# tests with the native addon REQUIRED, the wasm↔napi parity check, artifact
# verification, and the cross-package e2e through the @ruvnet/ruview
# metaharness. No publish step: publication stays CI-only via the release
# workflow and requires explicit maintainer approval (ADR-265).
name: ruview-kernel npm
on:
push:
branches: [main]
paths:
- 'v2/crates/ruview-kernel/**'
- 'v2/crates/ruview-kernel-napi/**'
- 'v2/crates/wifi-densepose-vitals/**'
- 'harness/ruview-kernel/**'
- 'harness/ruview/src/kernel.js'
- 'harness/ruview/test/kernel.test.mjs'
- '.github/workflows/ruview-kernel-npm.yml'
pull_request:
paths:
- 'v2/crates/ruview-kernel/**'
- 'v2/crates/ruview-kernel-napi/**'
- 'v2/crates/wifi-densepose-vitals/**'
- 'harness/ruview-kernel/**'
- 'harness/ruview/src/kernel.js'
- 'harness/ruview/test/kernel.test.mjs'
- '.github/workflows/ruview-kernel-npm.yml'
permissions:
contents: read
concurrency:
group: ruview-kernel-${{ github.ref }}
cancel-in-progress: true
jobs:
kernel:
name: kernel (${{ matrix.os }})
runs-on: ${{ matrix.os }}
timeout-minutes: 45
strategy:
fail-fast: false
matrix:
os: [ubuntu-latest, macos-latest, windows-latest]
defaults:
run:
shell: bash
steps:
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
with:
persist-credentials: false
# The v2 workspace lists submodule crates as members; cargo must be
# able to load every member manifest.
submodules: recursive
- name: Install Rust + wasm32 target
uses: dtolnay/rust-toolchain@4cda84d5c5c54efe2404f9d843567869ab1699d4
with:
targets: wasm32-unknown-unknown
# v2/rust-toolchain pins its own channel; the target must be installed
# for that toolchain, not only for the action's default `stable`.
- name: Add wasm32 target to the pinned v2 toolchain
working-directory: v2
run: rustup target add wasm32-unknown-unknown
- name: Cache cargo
uses: Swatinem/rust-cache@e18b497796c12c097a38f9edb9d0641fb99eee32
with:
workspaces: |
v2 -> target
v2/crates/ruview-kernel-napi -> target
- uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
with:
node-version: '22'
- name: Rust ABI tests
run: cargo test --manifest-path v2/Cargo.toml -p ruview-kernel --locked
- name: Build WASM + napi-rs artifacts from source
working-directory: harness/ruview-kernel
run: npm run build
- name: Verify artifacts (checksums, zero WASM imports, ABI)
working-directory: harness/ruview-kernel
run: node scripts/verify-artifacts.mjs --require-wasm
- name: Package tests (napi required)
working-directory: harness/ruview-kernel
env:
RUVIEW_KERNEL_REQUIRE_NAPI: '1'
run: npm test
- name: Doctor, self-test, and wasm↔napi parity (SYNTHETIC)
working-directory: harness/ruview-kernel
run: |
node bin/cli.js doctor --strict-napi
node bin/cli.js selftest --backend wasm
node bin/cli.js selftest --backend napi
node bin/cli.js parity
- name: Pack
working-directory: harness/ruview-kernel
run: |
npm pack --dry-run
# Relative destination: Windows runner temp paths contain
# backslashes that bash globbing would mangle.
mkdir -p dist-pack
npm pack --pack-destination dist-pack
- name: Cross-package e2e through the @ruvnet/ruview metaharness
working-directory: harness/ruview
run: |
npm ci --ignore-scripts
npm install --no-save --ignore-scripts ../ruview-kernel/dist-pack/ruvnet-ruview-kernel-*.tgz
node --test test/kernel.test.mjs
node bin/cli.js kernel --backend auto
- name: Upload built artifacts
if: matrix.os == 'ubuntu-latest'
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
with:
name: ruview-kernel-linux-x64
path: |
harness/ruview-kernel/wasm/ruview_kernel.wasm
harness/ruview-kernel/wasm/SHA256SUMS
harness/ruview-kernel/native/*.node
harness/ruview-kernel/native/SHA256SUMS
retention-days: 14
+8 -6
View File
@@ -104,8 +104,10 @@ jobs:
run: |
set -euo pipefail
case "${{ inputs.package }}" in
# ADR-283/325: brain + hosts + replay + guarded Spaces OAuth; no runtime deps.
harness/ruview) export UNPACKED_BUDGET=163840 ;;
# ADR-283/325/369-374: brain + hosts + replay + Spaces OAuth + SDK,
# flashing, training, doctor, device access, SSH hosts; ADR-375 console
# widget, HTTP transport, terminal UI; ADR-378 mod showcase; no runtime deps.
harness/ruview) export UNPACKED_BUDGET=458752 ;;
# ADR-285: CLI + MCP + reviewed brain + WASM-kernel adapter.
harness/homecore) export UNPACKED_BUDGET=180000 ;;
# ADR-264 O2: map-free tarball (was 188 kB with maps).
@@ -140,12 +142,12 @@ jobs:
npm i --no-fund --no-audit "$TGZ"
case "${{ inputs.package }}" in
harness/ruview)
./node_modules/.bin/ruview --version
./node_modules/.bin/ruview doctor
./node_modules/.bin/ruview guidance --topic homecore --query restore --limit 1 \
./node_modules/.bin/ruview-harness --version
./node_modules/.bin/ruview-harness doctor
./node_modules/.bin/ruview-harness guidance --topic homecore --query restore --limit 1 \
| grep -q '"homecore-runtime-restore"'
# the honesty gate must fail closed on empty input (ADR-263 F1)
if ./node_modules/.bin/ruview claim-check; then
if ./node_modules/.bin/ruview-harness claim-check; then
echo 'claim-check passed with no input — fail-open regression'; exit 1
fi
node --input-type=module -e "const m = await import('@ruvnet/ruview'); if (!m.TOOLS) process.exit(1);"
+3
View File
@@ -322,3 +322,6 @@ v2/crates/ruview-offaxis/pkg/
scripts/fal-visual-teacher/training-data/
scripts/fal-visual-teacher/training-data.zip
scripts/fal-visual-teacher/output/
# Trained adaptive classifier (auto-created at runtime; #2135)
/v2/data/adaptive_model.json
+21
View File
@@ -7,8 +7,29 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
## [Unreleased]
### Community batch (PRs #2100-#2157 by aepod, reviewed and integrated)
Integrated as one set: conflicts resolved, review findings fixed, validated on the integrated branch. Hardware figures cited by the PRs are the author's own live runs on ESP32-S3 nodes and were not independently reproduced here.
- **Security and honest status.** Refuse a routable HTTP bind when API auth is off (#2134). `--privacy-mode` now strips biometrics from REST, WebSocket and recordings, not only MQTT (#2125; the stripped-key list is an exact-match allowlist, so a new biometric field must be added to it). `/api/v1/status` waits for a real frame and unknown `--source` values are rejected (#2126; breaking for `--source macos|linux|windows`). A stale `/sensing/latest` reports `{"status":"stale"}` after 10 s (#2119). Measured stream fps and real WebSocket client counts (#2127). Absent nodes no longer count as one person (#2132). ESP32 frames with zero antennas or subcarriers are dropped (#2118). Dead nodes are evicted on a timer (#2120). Recordings and models stay under `--data-dir` (#2129). Boolean env flags accept 1/0/yes/no/on/off (#2108). Docker UDP ingest posture is explicit (#2107).
- **CSI pipeline.** The grid gate locks on the majority frame shape and `/health` reports a stalled pipeline (#2157; a partial fix, mixed 192/306 frame shapes still drop frames, so it does not close #1894). Phase is combined across nodes only when they declare a shared reference (#2156). Point-cloud and hardware parsers use the ADR-018 layout and accept more than 256 subcarriers (#2155; the bandwidth rule for counts above 64 is CLAIMED, inferred from the datasheet). Per-node adaptive training from real recordings (#2101, #2103); a run that fell back to the live buffer now reports `degraded: true`. Gaussian decay no longer compounds across calls (#2140).
- **MediaTek (ADR-266/267).** Legacy and unknown PPDU types in MTC1 (#2112), per-device ingest capped at 64 devices (#2113), a labelled amplitude heuristic and activity index (#2114; CLAIMED, not an occupancy estimate), an MT7981 bridge (#2115), and the OpenWrt WN586X3 patch set and guide (#2116; the vendored CSI patch is BSD-3-Clause-Clear, see `NOTICE`).
- **MQTT / Home Assistant.** Semantic states wired into the publisher, entities announced only when sourced, persisted client id (#2117). Existing Home Assistant installs keep retained discovery configs for the six removed entities until they are cleared manually.
- **Provisioning and firmware.** Wi-Fi password kept owner-only and out of argv and `--state` (#2153); state keyed by board MAC (#2152); `--dry-run` and a failed `--reset` leave state intact (#2111); OTA images confirmed by a bounded health check, which means only "reached the network and sent CSI" (#2100, ADR-379).
- **Ground truth and spatial.** Radar window adapter and auto-labeller with a session-disjoint split and a median count baseline; radar labels carry `evidence: CLAIMED` (#2143, #2144). Node self-survey, check-only (#2145, ADR-381). The `spatial.evidence.v1` export defaults to the CODE proof grade and MEASURED requires a reproducer (#2141, ADR-382). Loopback UDP tee (#2102, ADR-380).
- **Tooling.** `prove.sh` reports INCOMPLETE instead of PASS when a gate is skipped (#2130); ADR-018 capture scripts (#2131); API token on the ground-truth script (#2104); the UI asks for the API token instead of retrying forever (#2128); macOS redacted-link handling (#2105).
- **Not included:** #2106 (MacWifi.app, draft) and #2142 (MR60 probe change, needs a captured MR60BHA2 boot log). `v2/data/adaptive_model.json` (41.5% training accuracy) is no longer tracked (#2135).
### Added
- **`@ruvnet/ruview` 0.9.1: live sensing views inside Claude Code (ADR-377, ADR-378).** The package now ships `ruview-live`, a Claude Code mod. Type `/ruview` and a pane opens next to the conversation showing what your sensors see right now, refreshed automatically. It has three views, switched with keys `1` `2` `3` or with `/ruview overview|waterfall|radar`:
- **Overview**: the CSI nodes streaming to your machine (rate, packet loss, signal strength) and an optional 60 GHz radar kit (presence, distance, heart and breathing rates).
- **CSI waterfall**: the Wi-Fi signal itself, the strength of every subcarrier over time, drawn in colour and scrolling as frames arrive. The new `ruview esp32 --spectrum` option supplies the data.
- **Radar**: a top-down range fan with an animated ping out to the measured distance, plus heart, breathing and distance charts.
Every reading is labelled honestly: MEASURED or SYNTHETIC from the packets themselves; radar vitals as "device-reported, not validated"; range-only radar as "distance, not direction". The pane only reads; it never flashes or configures a device.
Verified on real hardware: an ESP32-C6 streaming CSI and a Seeed MR60BHA2 radar kit (MEASURED; reproducer `npx @ruvnet/ruview esp32 --seconds 4 --spectrum --json`). The package's size ceiling rises from 384 KiB to 448 KiB; it still has no runtime dependencies.
- **`wifi-densepose-sar` — coherent wideband RF tomography research crate (ADR-287).** New standalone leaf crate (the `nvsim` pattern; zero coupling to `wifi-densepose-hardware` or any real ingestion path) implementing the synthetic-aperture-radar reconstruction primitive a handheld through-wall RF imaging device would need — motivated by comparison against Applied Electrodynamics' "WaveSight" launch, and explicitly scoped below ADR-278's RISE/DiffRadar/GeRaF reproduction gates. Ships: (1) a stepped-frequency, multi-position complex forward measurement simulator (`y_{m,k} = Σ σ_j/R² · exp(-i·4π·f·R/c) + noise`, deterministic ChaCha20 seeding); (2) delay-and-sum backprojection reconstruction onto a 3D voxel grid, rayon-parallelized over voxels; (3) threshold + local-maximum point-cloud extraction; (4) closed-form range/cross-range resolution and antenna-pose coherence-budget formulas (`ΔR=c/2B`, `δ_CR≈λR/2L`, `Δp≤λ/8`) checked against the reconstruction's *actual* behavior in `tests/physics_validation.rs` rather than merely documented — forward-simulating two targets at controlled separations and proving they resolve or merge exactly where the formulas predict, and that reconstructed focus at a known target degrades as injected antenna-pose error grows. Every number is SYNTHETIC/L0 (ADR-282) — no real wideband RF hardware backs this crate; see the crate README and `docs/tutorials/coherent-rf-tomography-backprojection.md` for the full honesty boundary and a worked walkthrough. `focus_at_point` exploits the evenly-spaced-by-construction frequency sweep (an arithmetic progression in per-term phase) to evaluate each pose's phasor once and advance it by a fixed complex-multiply step per frequency instead of one `sin`/`cos` pair per frequency — **MEASURED ~4.4-4.5x faster** (criterion regression detection, p < 0.001) than the first-shipped direct-computation version, proven equivalent (not just faster) to an independently reimplemented reference across four sweep sizes and on-/off-target points. 25 tests (22 unit + 3 integration), 0 failed, clippy-clean; MEASURED backprojection throughput ~1.7-2.3M voxels/sec (criterion, 21 poses × 32 freq steps).
- **HOMECORE platform runtime completion — secure native/Wasmtime plugins, authenticated HAP IP, expanded Home Assistant APIs, durable restoration/migration, and voice protocols.** `homecore-server` now owns deterministic compiled-in native plugin registration plus explicitly configured, path-bounded, Ed25519 publisher-verified Wasm packages executed through Wasmtime with setup/state-change/teardown lifecycle; arbitrary native dynamic libraries remain intentionally unsupported. The optional HAP server implements persisted accessory identity and controller records, SRP-6a Pair-Setup M1–M6, X25519/Ed25519 Pair-Verify M1–M4, HKDF-SHA512/ChaCha20-Poly1305 record framing, authenticated/admin endpoint gates, replay/tamper closure, live entity synchronization, and paired-state `_hap._tcp` mDNS updates (45 focused tests; external Apple certification is not claimed). Startup restores device/entity registries and deterministic latest recorder states before plugins, and migration now atomically preserves forward-compatible device/config-entry fields. The HA-compatible surface adds events, templates, config checks, components, registries, history/logbook with SQL-enforced global response bounds, calendar/camera provider routes, and modern WebSocket negotiation while retaining a machine-readable limitations matrix for integration-specific behavior. Assist adds bounded PCM16, async STT/TTS contracts, an end-to-end speech pipeline, and an authenticated satellite session protocol; real deployments still provide the speech engines.
- **`ruview-unified` increment 3 — Gaussian update-loop completion, separable delay-Doppler, and property-tested boundary hardening.** (1) `GaussianMap::merge_overlapping` (ADR-275 step 5: mutual-Mahalanobis + semantic-compatibility dedup catching drift the insert-time gate misses) and lifetime-aware decay (`τ_eff = τ·(1+ln(1+lifetime/τ))` — confirmed structures outlive transients at equal nominal τ). (2) `delay_doppler_map` reimplemented separably (`O(B²S+S²B)`), proven equivalent to the direct reference to <1e-10 and **measured 8.3× faster** (520 µs vs 4.34 ms at 56×8). (3) `tests/security_boundaries.rs` — 8 `proptest` properties over the boundary surfaces (arbitrary values incl. NaN/±inf via `f64::from_bits`) that found and fixed three input-controlled defects: a BLE-CS phase-unwrap infinite loop on non-finite phases and an ~1e299-iteration loop on finite-huge phases (now O(1) modular unwrap + plausibility bound), and a subnormal Gaussian scale overflowing `1/σ²` to NaN density (now physical σ/occupancy bounds). (4) New criterion benches for all increment-2 hot paths (`to_canonical` 38 µs, `ble_cs_range` 481 ns, AoI planner 647 ns/200 regions, coherent fusion 1.5 µs/32 members, factorized pose 521 ns). ruview-unified now 98 tests (87 lib + 3 acceptance + 8 security), 0 failed, clippy-clean.
+22
View File
@@ -0,0 +1,22 @@
RuView third-party notices
This repository is licensed MIT OR Apache-2.0 (see LICENSE). The items below
are vendored third-party material that keeps its own licence and is not
covered by that dual licence.
firmware/openwrt-wn586x3/patches/1001-mtk-mt76-mt7915-csi-implement-csi-support-REBASED.patch
What: MediaTek's mt76 mt7915 CSI patch, rebased onto openwrt/mt76
eb567bc7 (the revision pinned by OpenWrt 24.10.8).
Author: Bo Jiao <Bo.Jiao@mediatek.com> (original author line kept in the
patch header); copyright MediaTek Inc.
Origin: mediatek/mtk-openwrt-feeds @ a15454c888f4f4144c50e33b5feef4f247c5f78b
Licence: BSD-3-Clause-Clear (SPDX identifier,
https://spdx.org/licenses/BSD-3-Clause-Clear.html). The patch
carries the SPDX tags and MediaTek copyright lines in its added
files; the licence text is not reproduced here.
Scope: A vendored patch applied to an OpenWrt/mt76 build. It is not part
of the Rust crates, is not compiled by the Cargo workspace, and is
not linked into any RuView binary.
See firmware/openwrt-wn586x3/README.md for the other patches in that
directory and the licence each one keeps.
+105 -18
View File
@@ -6,6 +6,27 @@
</a>
</p>
[![RuView: spaces become signals, with an animated conceptual RF field](assets/readme/hero-spatial-v3.svg)](docs/user-guide.md)
<table>
<tr>
<td align="center" width="33%"><a href="docs/user-guide.md"><img src="assets/readme/nav-start-spatial-v3.svg" width="180" alt="Start here" /></a><br/><a href="docs/user-guide.md">Start here</a></td>
<td align="center" width="33%"><a href="firmware/esp32-csi-node/README.md"><img src="assets/readme/nav-hardware-spatial-v3.svg" width="180" alt="Hardware" /></a><br/><a href="firmware/esp32-csi-node/README.md">Hardware</a></td>
<td align="center" width="33%"><a href="https://huggingface.co/ruvnet/wifi-densepose-pretrained"><img src="assets/readme/nav-models-spatial-v3.svg" width="180" alt="Models" /></a><br/><a href="https://huggingface.co/ruvnet/wifi-densepose-pretrained">Models</a></td>
</tr>
<tr>
<td align="center" width="33%"><a href="harness/ruview/README.md"><img src="assets/readme/nav-agents-spatial-v3.svg" width="180" alt="AI toolkit" /></a><br/><a href="harness/ruview/README.md">AI toolkit</a></td>
<td align="center" width="33%"><a href="docs/integrations/home-assistant.md"><img src="assets/readme/nav-home-spatial-v3.svg" width="180" alt="Smart home" /></a><br/><a href="docs/integrations/home-assistant.md">Smart home</a></td>
<td align="center" width="33%"><a href="https://cognitum.one/seed"><img src="assets/readme/nav-seed-spatial-v3.svg" width="180" alt="Cognitum Seed" /></a><br/><a href="https://cognitum.one/seed">Cognitum Seed</a></td>
</tr>
</table>
<p align="center">
<a href="https://ruos.cognitum.one">
<img src="assets/ruos-animated.svg" alt="RuView — WiFi becomes spatial awareness, with Ruflo coordination and a ruOS sensing workspace" width="100%">
</a>
</p>
## **See through walls with WiFi** ##
@@ -21,6 +42,8 @@ Works natively with the four major smart-home ecosystems: **[Home Assistant](doc
Every WiFi router already fills your space with radio waves. When people move, breathe, or even sit still, they disturb those waves in measurable ways. RuView captures these disturbances using Channel State Information (CSI) from low-cost ESP32 sensors and turns them into actionable data: who's there, what they're doing, and whether they're okay.
[![Explore RuView sensing, models and automation](assets/readme/header-features-spatial-v3.svg)](docs/user-guide.md)
**What it senses:**
- **Presence and occupancy** — detect people through walls, count them, track entries and exits
- **Vital signs** — breathing rate and heart rate, contactless, while sleeping or sitting
@@ -37,32 +60,75 @@ Every WiFi router already fills your space with radio waves. When people move, b
- **Governed evidence** — attach privacy policy, uncertainty, provenance, and witness records to sensing events
- **RuView MetaHarness** — use an AI operator to onboard, calibrate, train, verify, and check sensing claims
<!-- ruview-capability-cards -->
<table>
<tr>
<td width="50%"><a href="docs/user-guide.md"><img src="assets/readme/feature-presence-spatial-v3.svg" width="580" alt="Presence and motion: open documentation and current limitations" /></a><br/><a href="docs/user-guide.md">Presence and motion</a></td>
<td width="50%"><a href="v2/crates/wifi-densepose-signal/README.md"><img src="assets/readme/feature-vitals-spatial-v3.svg" width="580" alt="Breathing and vital signals: open documentation and current limitations" /></a><br/><a href="v2/crates/wifi-densepose-signal/README.md">Breathing and vital signals</a></td>
</tr>
<tr>
<td width="50%"><a href="docs/benchmarks/pose-estimation-cog.md"><img src="assets/readme/feature-pose-spatial-v3.svg" width="580" alt="Pose research and limitations: open documentation and current limitations" /></a><br/><a href="docs/benchmarks/pose-estimation-cog.md">Pose research and limitations</a></td>
<td width="50%"><a href="firmware/esp32-csi-node/README.md"><img src="assets/readme/feature-edge-spatial-v3.svg" width="580" alt="Edge sensor mesh: open documentation and current limitations" /></a><br/><a href="firmware/esp32-csi-node/README.md">Edge sensor mesh</a></td>
</tr>
<tr>
<td width="50%"><a href="docs/integrations/home-assistant.md"><img src="assets/readme/feature-automation-spatial-v3.svg" width="580" alt="Local automation: open documentation and current limitations" /></a><br/><a href="docs/integrations/home-assistant.md">Local automation</a></td>
<td width="50%"><a href="harness/ruview/README.md"><img src="assets/readme/feature-agents-spatial-v3.svg" width="580" alt="Guided AI operator: open documentation and current limitations" /></a><br/><a href="harness/ruview/README.md">Guided AI operator</a></td>
</tr>
</table>
<details>
<summary><strong>RuView MetaHarness</strong> — guided operation for humans and AI agents</summary>
The RuView-specific metaharness we created is published as [`@ruvnet/ruview`](harness/ruview/README.md). It provides source-cited guidance, guarded Claude Code/Codex agents, deterministic verification, an honesty check for accuracy claims, and an explicitly granted OAuth-only Cognitum Spaces read.
The RuView-specific metaharness we created is published as [`@ruvnet/ruview`](harness/ruview/README.md). It provides:
- source-cited guidance;
- device access for CSI nodes, radar and LiDAR;
- firmware flashing with boot evidence;
- one MCP server, usable over stdio or HTTP for ChatGPT, with a live console widget;
- a Claude Code mod;
- guarded Claude Code/Codex agents;
- deterministic verification;
- an honesty check for accuracy claims.
Companion packages: [`homecore`](harness/homecore/README.md) (Homecore developer metaharness) and [`@ruvnet/ruview-kernel`](harness/ruview-kernel/README.md) (the vitals pipeline as WASM).
```bash
# Check the local setup and get source-cited guidance
npx @ruvnet/ruview@0.4.0 doctor
npx @ruvnet/ruview@0.4.0 guidance --topic sensing --query "model loading"
npx @ruvnet/ruview@0.9.1 doctor
npx @ruvnet/ruview@0.9.1 guidance --topic sensing --query "model loading"
# Hardware: what is plugged in, live CSI from your nodes, a 60 GHz radar kit
npx @ruvnet/ruview@0.9.1 devices
npx @ruvnet/ruview@0.9.1 esp32 --watch # ESP32 + Realtek RAC1 nodes, live view
npx @ruvnet/ruview@0.9.1 esp32 --seconds 45 --analyze # live CSI through the vitals kernel
npx @ruvnet/ruview@0.9.1 mmwave --source esphome --host <kit-ip>
# Agents: MCP over stdio, or over HTTP for ChatGPT (token-protected, read tools only)
npx @ruvnet/ruview@0.9.1 mcp start
RUVIEW_MCP_GRANTS=device-access npx @ruvnet/ruview@0.9.1 mcp start --http
# Claude Code: a live sensing pane (/ruview), shipped in the package
npx @ruvnet/ruview@0.9.1 mod
# Run a read-only RuView agent through Codex
npx @ruvnet/ruview@0.4.0 agent run --host codex --repo . \
npx @ruvnet/ruview@0.9.1 agent run --host codex --repo . \
--prompt "Find the nearest tests and cite the source files"
# Search or verify the reviewed contributor brain
npx @ruvnet/ruview@0.4.0 brain search --query "calibration"
npx @ruvnet/ruview@0.4.0 brain verify --repo .
# Check claims, replay the deterministic proof, or expose the MCP server
npx @ruvnet/ruview@0.4.0 claim-check --file REPORT.md
npx @ruvnet/ruview@0.4.0 verify
npx @ruvnet/ruview@0.4.0 spaces
npx @ruvnet/ruview@0.4.0 mcp start
# Check claims, replay the deterministic proof, search the reviewed brain
npx @ruvnet/ruview@0.9.1 claim-check --file REPORT.md
npx @ruvnet/ruview@0.9.1 verify
npx @ruvnet/ruview@0.9.1 brain search --query "calibration"
```
Agent runs are read-only by default. Workspace writes require both `--allow-write` and `--confirm`; retrieved brain content is evidence, not authority.
**Safety:**
- Agent runs are read-only by default. Workspace writes require both `--allow-write` and `--confirm`.
- Hardware reads need the `device-access` grant.
- Flashing and calibration never run over the HTTP transport.
- Device-reported vitals are labelled as unvalidated.
- Retrieved brain content is evidence, not authority.
A single `npx ruview` package bundling all of this is ready (ADR-376). It is waiting on npm to release the unscoped name.
Full walkthrough: [user guide → RuView npm toolkit](docs/user-guide.md#ruview-npm-toolkit-ruvnetruview).
</details>
@@ -100,8 +166,8 @@ RuView turns ordinary WiFi into a contactless sensor. A $9 ESP32 board reads the
> | 🧱 **Through-wall sensing** | Fresnel-zone geometry + multipath modeling | Up to ~5 m, signal-dependent |
> | 🧠 **Edge intelligence** | **105-cog catalog** ([ADR-102](docs/adr/ADR-102-edge-module-registry.md)) live from `app-registry.json` — health, security, building, retail, industrial, research, AI, swarm, signal, network, and developer modules. Optional Cognitum Seed adds persistent vector store + kNN + witness chain | $140 total BOM |
> | 🎯 **Camera-free pre-training** | Self-supervised contrastive encoder, 12.2M training steps on 60K frames, shipped on Hugging Face | 84 s/epoch retrain on M4 Pro |
> | 📷 **Camera-supervised fine-tune** | MediaPipe + ESP32 CSI paired training, end-to-end Candle pipeline on RTX 5080 ([ADR-079](docs/adr/ADR-079-camera-supervised-pose-finetune.md)) | 2.1 s for 400 epochs (~5 ms/epoch) |
> | 📡 **Multi-frequency mesh** | Channel hopping across 6 bands, TDM slot scheduling ([ADR-029](docs/adr/ADR-029-multifrequency-mesh.md)) | 3× sensing bandwidth |
> | 📷 **Camera-supervised fine-tune** | MediaPipe + ESP32 CSI paired training, end-to-end Candle pipeline on RTX 5080 ([ADR-079](docs/adr/ADR-079-camera-ground-truth-training.md)) | 2.1 s for 400 epochs (~5 ms/epoch) |
> | 📡 **Multi-frequency mesh** | Channel hopping across 6 bands, TDM slot scheduling ([ADR-029](docs/adr/ADR-029-ruvsense-multistatic-sensing-mode.md)) | 3× sensing bandwidth |
> | 🌐 **3D point cloud fusion** | Camera depth (MiDaS) + WiFi CSI + mmWave radar → unified spatial model | 22 ms pipeline · 19K+ points/frame |
>
> Browse the full 105-module catalog (with practical descriptions, sizes, and difficulty) below in [🧩 Edge Module Catalog](#-edge-module-catalog), or visit [seed.cognitum.one/store](https://seed.cognitum.one/store).
@@ -114,8 +180,11 @@ RuView turns ordinary WiFi into a contactless sensor. A $9 ESP32 board reads the
```bash
# Option 1: Docker (simulated data, no hardware needed)
docker pull ruvnet/wifi-densepose:latest
docker run -p 3000:3000 ruvnet/wifi-densepose:latest
export RUVIEW_API_TOKEN=$(openssl rand -hex 32) # required; the container exits 64 without it
docker run -p 127.0.0.1:3000:3000 -e RUVIEW_API_TOKEN ruvnet/wifi-densepose:latest
# Open http://localhost:3000
# ESP32 nodes in Docker also need a UDP source guard (RUVIEW_UDP_ALLOW);
# see docs/user-guide.md "Receiving ESP32 frames in Docker".
# Option 2a: Live sensing with ESP32-S3 hardware ($9)
# Flash firmware, provision WiFi, and start sensing:
@@ -227,6 +296,8 @@ The separate **17-keypoint pose-estimation model** is now published at [`ruvnet/
### Results & proof
[![RuView visual guide: header-evidence](assets/readme/header-evidence-spatial-v3.svg)](docs/benchmarks/pose-estimation-cog.md)
See the measured benchmarks, witness records, and one-command reproducibility check.
<details>
@@ -281,6 +352,8 @@ number. The path to a first *reproducible* on-device baseline (PCK@20 ≥ 35%) i
## 🧩 Edge Module Catalog
[![RuView visual guide: header-edge](assets/readme/header-edge-spatial-v3.svg)](firmware/esp32-csi-node/README.md)
Add signed modules for health, security, buildings, industry, research, AI, and more.
<details>
@@ -455,6 +528,8 @@ Browse and install modules at [seed.cognitum.one/store](https://seed.cognitum.on
## 🔬 How It Works
[![RuView visual guide: overview](assets/readme/overview-spatial-v3.svg)](docs/user-guide.md)
WiFi routers flood every room with radio waves. When a person moves — or even breathes — those waves scatter differently. WiFi DensePose reads that scattering pattern and reconstructs what happened:
```
@@ -579,6 +654,8 @@ These scenarios exploit WiFi's ability to penetrate solid materials — concrete
## 🧠 Self-Learning WiFi AI
[![RuView visual guide: header-learning](assets/readme/header-learning-spatial-v3.svg)](docs/user-guide.md)
Learn compact room fingerprints from raw CSI and adapt the model to each environment.
<details>
@@ -694,7 +771,13 @@ claude --plugin-dir ./plugins/ruview
Verify the plugin structure: `bash plugins/ruview/scripts/smoke.sh`. Full details: [`plugins/ruview/README.md`](plugins/ruview/README.md).
For the portable RuView MetaHarness, use `npx @ruvnet/ruview@0.4.0`; the quick commands and fuller explanation are in the collapsed MetaHarness section near the top of this README and in [`harness/ruview/`](harness/ruview/README.md).
**Live sensing pane (Claude Code mod):** the same marketplace also lists `ruview-live`, a Claude Code mod.
- Install it with `/plugin install ruview-live@ruview`, or run `npx @ruvnet/ruview@0.9.1 mod` for its path inside the npm package.
- `/ruview` opens a pane beside the transcript with your CSI nodes and radar, refreshed on a timer, plus a one-line status.
- It is read-only.
- Mods are early access; if `/ruview` is missing, start Claude Code with `CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1`. See [ADR-377](docs/adr/ADR-377-ruview-live-claude-code-mod.md).
For the portable RuView MetaHarness, use `npx @ruvnet/ruview@0.9.1`. The quick commands and a fuller explanation are in the collapsed MetaHarness section near the top of this README, in [`harness/ruview/`](harness/ruview/README.md), and in the [user guide](docs/user-guide.md#ruview-npm-toolkit-ruvnetruview).
</details>
@@ -702,6 +785,8 @@ For the portable RuView MetaHarness, use `npx @ruvnet/ruview@0.4.0`; the quick c
## 📖 Documentation
[![RuView visual guide: header-docs](assets/readme/header-docs-spatial-v3.svg)](docs/user-guide.md)
Start with the user, build, and calibration guides; expand for the full reference map.
<details>
@@ -743,6 +828,8 @@ Start with the user, build, and calibration guides; expand for the full referenc
>
> Contributions and bug reports welcome at [Issues](https://github.com/ruvnet/RuView/issues).
[![Cognitum One: Ambient Intelligence at the edge of the physical world](assets/readme/cognitum-spatial-v3.svg)](https://cognitum.one)
## 📄 License
MIT License — see [LICENSE](LICENSE) for details.
+33
View File
@@ -10,6 +10,39 @@ func main() {
exit(1)
}
// One connected-link RSSI sample for the Rust adapter. Keep the legacy
// no-argument 10 Hz stream below unchanged for Python callers.
if CommandLine.arguments.contains("--scan-once") {
guard interface.powerOn(),
let channel = interface.wlanChannel(),
interface.rssiValue() < 0 else {
fputs("WiFi is not connected\n", stderr)
exit(1)
}
let sample: [String: Any] = [
// Always the connected link. macOS redacts ssid/bssid without
// Location Services; rssi/noise/channel stay real, so mark the
// sample as the connected link for the Rust adapter.
"connected": true,
"ssid": interface.ssid() ?? "",
"bssid": interface.bssid() ?? "00:00:00:00:00:00",
"channel": channel.channelNumber,
"rssi": interface.rssiValue(),
"noise": interface.noiseMeasurement(),
"timestamp": Date().timeIntervalSince1970,
"tx_rate": interface.transmitRate()
]
do {
let data = try JSONSerialization.data(withJSONObject: sample, options: [.sortedKeys])
FileHandle.standardOutput.write(data)
FileHandle.standardOutput.write(Data([0x0a]))
} catch {
fputs("Could not encode WiFi sample: \(error)\n", stderr)
exit(1)
}
return
}
// Flush stdout automatically to prevent buffering issues with Python subprocess
setbuf(stdout, nil)
+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="304" viewBox="0 0 1200 304" role="img" aria-labelledby="title desc"><title id="title">Explore Cognitum One</title><desc id="desc">Ambient Intelligence at the edge of the physical world. Open cognitum.one.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="303" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="44" y="42" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">COGNITUM ONE</text><text x="44" y="99" font-size="42" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Ambient Intelligence.</text><text x="44" y="147" font-size="25" fill="#78cfff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >At the edge of the physical world.</text><text x="44" y="190" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Explore the platform behind the ecosystem.</text><rect x="44" y="224" width="256" height="46" rx="11" fill="url(#accent)"/><text x="65" y="254" font-size="18" fill="#071124" text-anchor="start" font-weight="500" >Explore Cognitum</text><path d="M263 247H284M277 240L284 247L277 254" fill="none" stroke="#071124" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="947" cy="151" r="173" fill="url(#halo)"/><g transform="translate(945 173) scale(1.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(945 145) scale(1.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 -38L31 -20V18L0 38L-31 18V-20ZM-31 -20L0 -1L31 -20M0 -1V38" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><path d="M-15 1L-5 11L17 -12" fill="none" stroke="#438fff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/></g></g><ellipse cx="945" cy="150" rx="140" ry="30" transform="rotate(-25 945 150)" fill="none" stroke="#437cb2" stroke-dasharray="5 18" class="flow"/><ellipse cx="945" cy="150" rx="160" ry="43" transform="rotate(-25 945 150)" fill="none" stroke="#437cb2" stroke-dasharray="5 18" class="flow"/><ellipse cx="945" cy="150" rx="180" ry="56" transform="rotate(-25 945 150)" fill="none" stroke="#437cb2" stroke-dasharray="5 18" class="flow"/></svg>

After

Width:  |  Height:  |  Size: 5.0 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="304" viewBox="0 0 1200 304" role="img" aria-labelledby="title desc"><title id="title">Explore Cognitum One</title><desc id="desc">Ambient Intelligence at the edge of the physical world. Open cognitum.one.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="303" rx="18" fill="url(#wash)" stroke="#245360"/><text x="44" y="42" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">COGNITUM ONE</text><text x="44" y="99" font-size="42" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Ambient Intelligence.</text><text x="44" y="147" font-size="25" fill="#72eee0" text-anchor="start" font-weight="650" letter-spacing="-0.6" >At the edge of the physical world.</text><text x="44" y="190" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Explore the platform behind the ecosystem.</text><rect x="44" y="224" width="256" height="46" rx="11" fill="url(#accent)"/><text x="65" y="254" font-size="18" fill="#041117" text-anchor="start" font-weight="500" >Explore Cognitum</text><path d="M263 247H284M277 240L284 247L277 254" fill="none" stroke="#041117" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="947" cy="151" r="173" fill="url(#halo)"/><g transform="translate(945 173) scale(1.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(945 145) scale(1.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 -38L31 -20V18L0 38L-31 18V-20ZM-31 -20L0 -1L31 -20M0 -1V38" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><path d="M-15 1L-5 11L17 -12" fill="none" stroke="#70b6ff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/></g></g><ellipse cx="945" cy="150" rx="140" ry="30" transform="rotate(-25 945 150)" fill="none" stroke="#3c8897" stroke-dasharray="5 18" class="flow"/><ellipse cx="945" cy="150" rx="160" ry="43" transform="rotate(-25 945 150)" fill="none" stroke="#3c8897" stroke-dasharray="5 18" class="flow"/><ellipse cx="945" cy="150" rx="180" ry="56" transform="rotate(-25 945 150)" fill="none" stroke="#3c8897" stroke-dasharray="5 18" class="flow"/></svg>

After

Width:  |  Height:  |  Size: 4.7 KiB

@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">A guided AI operator</title><desc id="desc">Source-cited setup and verification. CLI, MCP and agent workflows. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="30" y="39" font-size="11" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">RUVIEW / 06</text><text x="30" y="94" font-size="29" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >A guided AI operator</text><text x="30" y="136" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Source-cited setup and verification.</text><text x="30" y="164" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >CLI, MCP and agent workflows.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 0L0 -33" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="0" cy="-33" r="7" stroke="#78cfff" fill="none" class="breathe"/><path d="M0 0L-35 23" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="-35" cy="23" r="7" stroke="#78cfff" fill="none" class="breathe"/><path d="M0 0L35 23" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="35" cy="23" r="7" stroke="#78cfff" fill="none" class="breathe"/><circle cx="0" cy="0" r="15" stroke="#78cfff" fill="none" /><path d="M-6 0L-1 5L8 -6" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><path d="M30 204H550" fill="none" stroke="#2d4e77" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#78cfff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">TOOLKIT GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 5.2 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">A guided AI operator</title><desc id="desc">Source-cited setup and verification. CLI, MCP and agent workflows. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#245360"/><text x="30" y="39" font-size="11" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">RUVIEW / 06</text><text x="30" y="94" font-size="29" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >A guided AI operator</text><text x="30" y="136" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Source-cited setup and verification.</text><text x="30" y="164" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >CLI, MCP and agent workflows.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 0L0 -33" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="0" cy="-33" r="7" stroke="#72eee0" fill="none" class="breathe"/><path d="M0 0L-35 23" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="-35" cy="23" r="7" stroke="#72eee0" fill="none" class="breathe"/><path d="M0 0L35 23" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="35" cy="23" r="7" stroke="#72eee0" fill="none" class="breathe"/><circle cx="0" cy="0" r="15" stroke="#72eee0" fill="none" /><path d="M-6 0L-1 5L8 -6" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><path d="M30 204H550" fill="none" stroke="#245360" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#72eee0" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">TOOLKIT GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 5.0 KiB

@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Local automation</title><desc id="desc">Connect rooms to your smart home. Home Assistant, MQTT and Matter. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="30" y="39" font-size="11" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">RUVIEW / 05</text><text x="30" y="94" font-size="29" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Local automation</text><text x="30" y="136" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Connect rooms to your smart home.</text><text x="30" y="164" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Home Assistant, MQTT and Matter.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-37 -3L0 -34L37 -3M-28 -10V32H28V-10M-9 32V8H9V32" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-9" r="5" stroke="#438fff" fill="none" class="breathe"/></g></g><path d="M30 204H550" fill="none" stroke="#2d4e77" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#78cfff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">INTEGRATION GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.7 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Local automation</title><desc id="desc">Connect rooms to your smart home. Home Assistant, MQTT and Matter. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#245360"/><text x="30" y="39" font-size="11" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">RUVIEW / 05</text><text x="30" y="94" font-size="29" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Local automation</text><text x="30" y="136" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Connect rooms to your smart home.</text><text x="30" y="164" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Home Assistant, MQTT and Matter.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-37 -3L0 -34L37 -3M-28 -10V32H28V-10M-9 32V8H9V32" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-9" r="5" stroke="#70b6ff" fill="none" class="breathe"/></g></g><path d="M30 204H550" fill="none" stroke="#245360" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#72eee0" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">INTEGRATION GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.4 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Edge sensor mesh</title><desc id="desc">Provision, capture and calibrate. Explore ESP32 firmware and runbooks. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="30" y="39" font-size="11" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">RUVIEW / 04</text><text x="30" y="94" font-size="29" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Edge sensor mesh</text><text x="30" y="136" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Provision, capture and calibrate.</text><text x="30" y="164" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Explore ESP32 firmware and runbooks.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><rect x="-25" y="-25" width="50" height="50" rx="7" fill="#0c2242" stroke="#78cfff"/><rect x="-13" y="-13" width="26" height="26" rx="3" fill="none" stroke="#438fff"/><path d="M-16 -36V-25M-16 25V36M-36 -16H-25M25 -16H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M0 -36V-25M0 25V36M-36 0H-25M25 0H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M16 -36V-25M16 25V36M-36 16H-25M25 16H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="0" r="4" stroke="#78cfff" fill="#78cfff" class="breathe"/></g></g><path d="M30 204H550" fill="none" stroke="#2d4e77" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#78cfff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">HARDWARE GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 5.1 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Edge sensor mesh</title><desc id="desc">Provision, capture and calibrate. Explore ESP32 firmware and runbooks. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#245360"/><text x="30" y="39" font-size="11" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">RUVIEW / 04</text><text x="30" y="94" font-size="29" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Edge sensor mesh</text><text x="30" y="136" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Provision, capture and calibrate.</text><text x="30" y="164" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Explore ESP32 firmware and runbooks.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><rect x="-25" y="-25" width="50" height="50" rx="7" fill="#0a2630" stroke="#72eee0"/><rect x="-13" y="-13" width="26" height="26" rx="3" fill="none" stroke="#70b6ff"/><path d="M-16 -36V-25M-16 25V36M-36 -16H-25M25 -16H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M0 -36V-25M0 25V36M-36 0H-25M25 0H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M16 -36V-25M16 25V36M-36 16H-25M25 16H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="0" r="4" stroke="#72eee0" fill="#72eee0" class="breathe"/></g></g><path d="M30 204H550" fill="none" stroke="#245360" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#72eee0" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">HARDWARE GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.9 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Pose research</title><desc id="desc">Train and compare body keypoints. Live accuracy differs from benchmarks. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="30" y="39" font-size="11" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">RUVIEW / 03</text><text x="30" y="94" font-size="29" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Pose research</text><text x="30" y="136" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Train and compare body keypoints.</text><text x="30" y="164" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Live accuracy differs from benchmarks.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><circle cx="0" cy="-28" r="8" stroke="#b8f36c" fill="none" /><path d="M0 -20V12M-24 -6L0 -14L24 -6M0 12L-18 38M0 12L18 38" fill="none" stroke="#b8f36c" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-14" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="0" cy="12" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="-24" cy="-6" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="24" cy="-6" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="-18" cy="38" r="3" stroke="#ffb45e" fill="#071124" class="breathe"/><circle cx="18" cy="38" r="3" stroke="#ffb45e" fill="#071124" class="breathe"/></g></g><path d="M30 204H550" fill="none" stroke="#2d4e77" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#78cfff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">READ THE EVIDENCE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 5.1 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Pose research</title><desc id="desc">Train and compare body keypoints. Live accuracy differs from benchmarks. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#245360"/><text x="30" y="39" font-size="11" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">RUVIEW / 03</text><text x="30" y="94" font-size="29" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Pose research</text><text x="30" y="136" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Train and compare body keypoints.</text><text x="30" y="164" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Live accuracy differs from benchmarks.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><circle cx="0" cy="-28" r="8" stroke="#72eee0" fill="none" /><path d="M0 -20V12M-24 -6L0 -14L24 -6M0 12L-18 38M0 12L18 38" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-14" r="3" stroke="#70b6ff" fill="#041117" class="breathe"/><circle cx="0" cy="12" r="3" stroke="#70b6ff" fill="#041117" class="breathe"/><circle cx="-24" cy="-6" r="3" stroke="#70b6ff" fill="#041117" class="breathe"/><circle cx="24" cy="-6" r="3" stroke="#70b6ff" fill="#041117" class="breathe"/><circle cx="-18" cy="38" r="3" stroke="#70b6ff" fill="#041117" class="breathe"/><circle cx="18" cy="38" r="3" stroke="#70b6ff" fill="#041117" class="breathe"/></g></g><path d="M30 204H550" fill="none" stroke="#245360" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#72eee0" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">READ THE EVIDENCE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.9 KiB

@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Presence &amp; motion</title><desc id="desc">Explore changes in room occupancy. Inspect motion features from CSI. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="30" y="39" font-size="11" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">RUVIEW / 01</text><text x="30" y="94" font-size="29" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Presence &amp; motion</text><text x="30" y="136" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Explore changes in room occupancy.</text><text x="30" y="164" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Inspect motion features from CSI.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><circle cx="0" cy="-28" r="8" stroke="#b8f36c" fill="none" /><path d="M0 -20V12M-24 -6L0 -14L24 -6M0 12L-18 38M0 12L18 38" fill="none" stroke="#b8f36c" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-14" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="0" cy="12" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="-24" cy="-6" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="24" cy="-6" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><ellipse cx="0" cy="14" rx="40" ry="13" fill="none" stroke="#438fff" stroke-dasharray="3 8" class="flow"/></g></g><path d="M30 204H550" fill="none" stroke="#2d4e77" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#78cfff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">SENSING GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 5.1 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Presence &amp; motion</title><desc id="desc">Explore changes in room occupancy. Inspect motion features from CSI. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#245360"/><text x="30" y="39" font-size="11" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">RUVIEW / 01</text><text x="30" y="94" font-size="29" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Presence &amp; motion</text><text x="30" y="136" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Explore changes in room occupancy.</text><text x="30" y="164" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Inspect motion features from CSI.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><circle cx="0" cy="-28" r="8" stroke="#72eee0" fill="none" /><path d="M0 -20V12M-24 -6L0 -14L24 -6M0 12L-18 38M0 12L18 38" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><ellipse cx="0" cy="14" rx="40" ry="13" fill="none" stroke="#70b6ff" stroke-dasharray="3 8" class="flow"/></g></g><path d="M30 204H550" fill="none" stroke="#245360" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#72eee0" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">SENSING GUIDE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.5 KiB

@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Breathing &amp; vital signals</title><desc id="desc">Inspect phase-derived signal trends. Validate estimates on your hardware. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="30" y="39" font-size="11" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">RUVIEW / 02</text><text x="30" y="94" font-size="29" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Breathing &amp; vital signals</text><text x="30" y="136" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Inspect phase-derived signal trends.</text><text x="30" y="164" font-size="16" fill="#b0c2db" text-anchor="start" font-weight="500" >Validate estimates on your hardware.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-42 6H-27L-18 -10L-9 25L3 -25L14 10L24 -3L31 6H43" fill="none" stroke="#b8f36c" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/><path d="M-42 31H43" fill="none" stroke="#233f63" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /></g></g><path d="M30 204H550" fill="none" stroke="#2d4e77" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#78cfff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">SIGNAL PIPELINE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.8 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="580" height="276" viewBox="0 0 580 276" role="img" aria-labelledby="title desc"><title id="title">Breathing &amp; vital signals</title><desc id="desc">Inspect phase-derived signal trends. Validate estimates on your hardware. Conceptual animation; follow the linked source for capability limits.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="579" height="275" rx="18" fill="url(#wash)" stroke="#245360"/><text x="30" y="39" font-size="11" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">RUVIEW / 02</text><text x="30" y="94" font-size="29" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Breathing &amp; vital signals</text><text x="30" y="136" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Inspect phase-derived signal trends.</text><text x="30" y="164" font-size="16" fill="#9abac5" text-anchor="start" font-weight="500" >Validate estimates on your hardware.</text><circle cx="465" cy="170" r="110" fill="url(#halo)"/><g transform="translate(480 161) scale(0.53)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(474 155) scale(0.68)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-42 6H-27L-18 -10L-9 25L3 -25L14 10L24 -3L31 6H43" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/><path d="M-42 31H43" fill="none" stroke="#214451" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /></g></g><path d="M30 204H550" fill="none" stroke="#245360" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M30 204H550" fill="none" stroke="#72eee0" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow" opacity=".5"/><text x="30" y="245" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">SIGNAL PIPELINE</text><path d="M510 239H540M532 231L540 239L532 247" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.5 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Choose your next step</title><desc id="desc">Animated section heading. GUIDES / HARDWARE / INTEGRATIONS</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 0L0 -33" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="0" cy="-33" r="7" stroke="#78cfff" fill="none" class="breathe"/><path d="M0 0L-35 23" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="-35" cy="23" r="7" stroke="#78cfff" fill="none" class="breathe"/><path d="M0 0L35 23" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="35" cy="23" r="7" stroke="#78cfff" fill="none" class="breathe"/><circle cx="0" cy="0" r="15" stroke="#78cfff" fill="none" /><path d="M-6 0L-1 5L8 -6" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="140" y="63" font-size="34" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Choose your next step</text><text x="142" y="99" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">GUIDES / HARDWARE / INTEGRATIONS</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.5 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Choose your next step</title><desc id="desc">Animated section heading. GUIDES / HARDWARE / INTEGRATIONS</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 0L0 -33" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="0" cy="-33" r="7" stroke="#72eee0" fill="none" class="breathe"/><path d="M0 0L-35 23" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="-35" cy="23" r="7" stroke="#72eee0" fill="none" class="breathe"/><path d="M0 0L35 23" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="35" cy="23" r="7" stroke="#72eee0" fill="none" class="breathe"/><circle cx="0" cy="0" r="15" stroke="#72eee0" fill="none" /><path d="M-6 0L-1 5L8 -6" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="140" y="63" font-size="34" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Choose your next step</text><text x="142" y="99" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">GUIDES / HARDWARE / INTEGRATIONS</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.3 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Intelligence at the edge</title><desc id="desc">Animated section heading. SENSORS / FIRMWARE / MODULES</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><rect x="-25" y="-25" width="50" height="50" rx="7" fill="#0c2242" stroke="#78cfff"/><rect x="-13" y="-13" width="26" height="26" rx="3" fill="none" stroke="#438fff"/><path d="M-16 -36V-25M-16 25V36M-36 -16H-25M25 -16H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M0 -36V-25M0 25V36M-36 0H-25M25 0H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M16 -36V-25M16 25V36M-36 16H-25M25 16H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="0" r="4" stroke="#78cfff" fill="#78cfff" class="breathe"/></g></g><text x="140" y="63" font-size="34" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Intelligence at the edge</text><text x="142" y="99" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">SENSORS / FIRMWARE / MODULES</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.4 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Intelligence at the edge</title><desc id="desc">Animated section heading. SENSORS / FIRMWARE / MODULES</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><rect x="-25" y="-25" width="50" height="50" rx="7" fill="#0a2630" stroke="#72eee0"/><rect x="-13" y="-13" width="26" height="26" rx="3" fill="none" stroke="#70b6ff"/><path d="M-16 -36V-25M-16 25V36M-36 -16H-25M25 -16H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M0 -36V-25M0 25V36M-36 0H-25M25 0H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M16 -36V-25M16 25V36M-36 16H-25M25 16H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="0" r="4" stroke="#72eee0" fill="#72eee0" class="breathe"/></g></g><text x="140" y="63" font-size="34" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Intelligence at the edge</text><text x="142" y="99" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">SENSORS / FIRMWARE / MODULES</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.1 KiB

@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Evidence before accuracy claims</title><desc id="desc">Animated section heading. BENCHMARKS / REPRODUCERS / LIMITS</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 -38L31 -20V18L0 38L-31 18V-20ZM-31 -20L0 -1L31 -20M0 -1V38" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><path d="M-15 1L-5 11L17 -12" fill="none" stroke="#438fff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/></g></g><text x="140" y="63" font-size="34" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Evidence before accuracy claims</text><text x="142" y="99" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">BENCHMARKS / REPRODUCERS / LIMITS</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.0 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Evidence before accuracy claims</title><desc id="desc">Animated section heading. BENCHMARKS / REPRODUCERS / LIMITS</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 -38L31 -20V18L0 38L-31 18V-20ZM-31 -20L0 -1L31 -20M0 -1V38" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><path d="M-15 1L-5 11L17 -12" fill="none" stroke="#70b6ff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/></g></g><text x="140" y="63" font-size="34" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Evidence before accuracy claims</text><text x="142" y="99" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">BENCHMARKS / REPRODUCERS / LIMITS</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 3.8 KiB

@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Explore the capabilities</title><desc id="desc">Animated section heading. SENSING / MODELS / AUTOMATION</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><circle cx="0" cy="-28" r="8" stroke="#b8f36c" fill="none" /><path d="M0 -20V12M-24 -6L0 -14L24 -6M0 12L-18 38M0 12L18 38" fill="none" stroke="#b8f36c" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-14" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="0" cy="12" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="-24" cy="-6" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><circle cx="24" cy="-6" r="2.3" stroke="#ffb45e" fill="#ffb45e" class="spark"/><ellipse cx="0" cy="14" rx="40" ry="13" fill="none" stroke="#438fff" stroke-dasharray="3 8" class="flow"/></g></g><text x="140" y="63" font-size="34" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Explore the capabilities</text><text x="142" y="99" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">SENSING / MODELS / AUTOMATION</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.4 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">Explore the capabilities</title><desc id="desc">Animated section heading. SENSING / MODELS / AUTOMATION</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><circle cx="0" cy="-28" r="8" stroke="#72eee0" fill="none" /><path d="M0 -20V12M-24 -6L0 -14L24 -6M0 12L-18 38M0 12L18 38" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><ellipse cx="0" cy="14" rx="40" ry="13" fill="none" stroke="#70b6ff" stroke-dasharray="3 8" class="flow"/></g></g><text x="140" y="63" font-size="34" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Explore the capabilities</text><text x="142" y="99" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">SENSING / MODELS / AUTOMATION</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 3.8 KiB

@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">From recordings to models</title><desc id="desc">Animated section heading. CAPTURE / TRAIN / EVALUATE</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-42 6H-27L-18 -10L-9 25L3 -25L14 10L24 -3L31 6H43" fill="none" stroke="#b8f36c" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/><path d="M-42 31H43" fill="none" stroke="#233f63" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="140" y="63" font-size="34" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >From recordings to models</text><text x="142" y="99" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">CAPTURE / TRAIN / EVALUATE</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 4.0 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="148" viewBox="0 0 1200 148" role="img" aria-labelledby="title desc"><title id="title">From recordings to models</title><desc id="desc">Animated section heading. CAPTURE / TRAIN / EVALUATE</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="147" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(76 75) scale(0.75)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-42 6H-27L-18 -10L-9 25L3 -25L14 10L24 -3L31 6H43" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/><path d="M-42 31H43" fill="none" stroke="#214451" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="140" y="63" font-size="34" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >From recordings to models</text><text x="142" y="99" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">CAPTURE / TRAIN / EVALUATE</text><g transform="translate(1012 84) scale(0.55)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><path d="M1140 70H1160M1152 62L1160 70L1152 78" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /></svg>

After

Width:  |  Height:  |  Size: 3.7 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="450" viewBox="0 0 1200 450" role="img" aria-labelledby="title desc"><title id="title">RuView: spaces become signals</title><desc id="desc">Animated conceptual RF field and room geometry. Explore the RuView user guide. This illustration is not a sensor measurement.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="449" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="48" y="44" font-size="13" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">RUVIEW / COGNITUM ECOSYSTEM</text><text x="48" y="121" font-size="76" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >RuView</text><text x="48" y="184" font-size="38" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Spaces become signals.</text><text x="48" y="234" font-size="32" fill="#78cfff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Signals become understanding.</text><text x="50" y="279" font-size="18" fill="#b0c2db" text-anchor="start" font-weight="500" >Explore RF sensing, edge inference and local automation.</text><rect x="48" y="326" width="270" height="52" rx="12" fill="url(#accent)"/><text x="72" y="359" font-size="20" fill="#071124" text-anchor="start" font-weight="500" >Explore the guide</text><path d="M272 352H292M285 345L292 352L285 359" fill="none" stroke="#071124" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><text x="49" y="419" font-size="11" fill="#b0c2db" text-anchor="start" font-weight="500" class="mono">RF SENSING / EDGE MODELS / LOCAL ACTION</text><circle cx="950" cy="235" r="235" fill="url(#halo)"/><path d="M750 282L610 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M790 282L690 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M830 282L770 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M870 282L850 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M910 282L930 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M950 282L1010 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M990 282L1090 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M1030 282L1170 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M1070 282L1250 443" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 306H1168" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 336H1168" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 372H1168" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 418H1168" fill="none" stroke="#162f50" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M803 157L1008 106L1137 177L930 236ZM803 157V314L930 391L1137 324V177M930 236V391M1008 106V268L803 314M1008 268L1137 324" fill="none" stroke="#5490c5" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g><g transform="translate(0 80)"><g class="scan"><path d="M803 157L1008 106L1137 177L930 236Z" fill="#78cfff" fill-opacity=".1" stroke="#78cfff" stroke-opacity=".6"/></g></g><g transform="translate(974 282)"><g class="orbit"><ellipse rx="135" ry="36" fill="none" stroke="#438fff" stroke-dasharray="3 11" opacity=".4"/><circle cx="135" cy="0" r="4" fill="#9ee7ff"/></g></g><g transform="translate(974 312)"><ellipse rx="140" ry="44" fill="none" stroke="#438fff" class="wave" style="animation-delay:-0.0s"/></g><g transform="translate(974 312)"><ellipse rx="140" ry="44" fill="none" stroke="#438fff" class="wave" style="animation-delay:-1.3s"/></g><g transform="translate(974 312)"><ellipse rx="140" ry="44" fill="none" stroke="#438fff" class="wave" style="animation-delay:-2.6s"/></g><circle cx="849" cy="198" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-0.0s"/><circle cx="893" cy="198" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-0.5s"/><circle cx="937" cy="198" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-1.0s"/><circle cx="981" cy="198" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-1.5s"/><circle cx="1025" cy="198" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-2.0s"/><circle cx="1069" cy="198" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-2.5s"/><circle cx="849" cy="233" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-3.0s"/><circle cx="893" cy="233" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-0.0s"/><circle cx="937" cy="233" r="1.7" stroke="#438fff" fill="#438fff" class="spark" style="animation-delay:-0.5s"/><circle cx="981" cy="233" r="1.7" stroke="#43Line truncated

After

Width:  |  Height:  |  Size: 12 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="450" viewBox="0 0 1200 450" role="img" aria-labelledby="title desc"><title id="title">RuView: spaces become signals</title><desc id="desc">Animated conceptual RF field and room geometry. Explore the RuView user guide. This illustration is not a sensor measurement.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="449" rx="18" fill="url(#wash)" stroke="#245360"/><text x="48" y="44" font-size="13" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">RUVIEW / COGNITUM ECOSYSTEM</text><text x="48" y="121" font-size="76" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >RuView</text><text x="48" y="184" font-size="38" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Spaces become signals.</text><text x="48" y="234" font-size="32" fill="#72eee0" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Signals become understanding.</text><text x="50" y="279" font-size="18" fill="#9abac5" text-anchor="start" font-weight="500" >Explore RF sensing, edge inference and local automation.</text><rect x="48" y="326" width="270" height="52" rx="12" fill="url(#accent)"/><text x="72" y="359" font-size="20" fill="#041117" text-anchor="start" font-weight="500" >Explore the guide</text><path d="M272 352H292M285 345L292 352L285 359" fill="none" stroke="#041117" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><text x="49" y="419" font-size="11" fill="#9abac5" text-anchor="start" font-weight="500" class="mono">RF SENSING / EDGE MODELS / LOCAL ACTION</text><circle cx="950" cy="235" r="235" fill="url(#halo)"/><path d="M750 282L610 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M790 282L690 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M830 282L770 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M870 282L850 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M910 282L930 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M950 282L1010 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M990 282L1090 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M1030 282L1170 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M1070 282L1250 443" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 306H1168" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 336H1168" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 372H1168" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><path d="M680 418H1168" fill="none" stroke="#12343d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M803 157L1008 106L1137 177L930 236ZM803 157V314L930 391L1137 324V177M930 236V391M1008 106V268L803 314M1008 268L1137 324" fill="none" stroke="#4b9ba4" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g><g transform="translate(0 80)"><g class="scan"><path d="M803 157L1008 106L1137 177L930 236Z" fill="#72eee0" fill-opacity=".1" stroke="#72eee0" stroke-opacity=".6"/></g></g><g transform="translate(974 282)"><g class="orbit"><ellipse rx="135" ry="36" fill="none" stroke="#70b6ff" stroke-dasharray="3 11" opacity=".4"/><circle cx="135" cy="0" r="4" fill="#91f6df"/></g></g><circle cx="849" cy="198" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-0.0s"/><circle cx="893" cy="198" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-0.5s"/><circle cx="937" cy="198" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-1.0s"/><circle cx="981" cy="198" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-1.5s"/><circle cx="1025" cy="198" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-2.0s"/><circle cx="1069" cy="198" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-2.5s"/><circle cx="849" cy="233" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-3.0s"/><circle cx="893" cy="233" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-0.0s"/><circle cx="937" cy="233" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-0.5s"/><circle cx="981" cy="233" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-1.0s"/><circle cx="1025" cy="233" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-1.5s"/><circle cx="1069" cy="233" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-2.0s"/><circle cx="849" cy="268" r="1.7" stroke="#70b6ff" fill="#70b6ff" class="spark" style="animation-delay:-2.5s"/><circlLine truncated

After

Width:  |  Height:  |  Size: 11 KiB

+147
View File
@@ -0,0 +1,147 @@
{
"schema_version": 1,
"style": "RuView navy, electric blue, lime and amber",
"motion": "Decorative CSS; prefers-reduced-motion supported; no script or remote dependencies",
"assets": [
{
"file": "assets/readme/hero-spatial-v3.svg",
"title": "RuView: spaces become signals",
"target": "docs/user-guide.md",
"width": 1200,
"height": 450
},
{
"file": "assets/readme/nav-start-spatial-v3.svg",
"title": "Start here",
"target": "docs/user-guide.md",
"width": 180,
"height": 116
},
{
"file": "assets/readme/nav-hardware-spatial-v3.svg",
"title": "Hardware",
"target": "firmware/esp32-csi-node/README.md",
"width": 180,
"height": 116
},
{
"file": "assets/readme/nav-models-spatial-v3.svg",
"title": "Models",
"target": "https://huggingface.co/ruvnet/wifi-densepose-pretrained",
"width": 180,
"height": 116
},
{
"file": "assets/readme/nav-agents-spatial-v3.svg",
"title": "AI toolkit",
"target": "harness/ruview/README.md",
"width": 180,
"height": 116
},
{
"file": "assets/readme/nav-home-spatial-v3.svg",
"title": "Smart home",
"target": "docs/integrations/home-assistant.md",
"width": 180,
"height": 116
},
{
"file": "assets/readme/nav-seed-spatial-v3.svg",
"title": "Cognitum",
"target": "https://cognitum.one/seed",
"width": 180,
"height": 116
},
{
"file": "assets/readme/feature-presence-spatial-v3.svg",
"title": "Presence & motion",
"target": "docs/user-guide.md",
"width": 580,
"height": 276
},
{
"file": "assets/readme/feature-vitals-spatial-v3.svg",
"title": "Breathing & vital signals",
"target": "v2/crates/wifi-densepose-signal/README.md",
"width": 580,
"height": 276
},
{
"file": "assets/readme/feature-pose-spatial-v3.svg",
"title": "Pose research",
"target": "docs/benchmarks/pose-estimation-cog.md",
"width": 580,
"height": 276
},
{
"file": "assets/readme/feature-edge-spatial-v3.svg",
"title": "Edge sensor mesh",
"target": "firmware/esp32-csi-node/README.md",
"width": 580,
"height": 276
},
{
"file": "assets/readme/feature-automation-spatial-v3.svg",
"title": "Local automation",
"target": "docs/integrations/home-assistant.md",
"width": 580,
"height": 276
},
{
"file": "assets/readme/feature-agents-spatial-v3.svg",
"title": "A guided AI operator",
"target": "harness/ruview/README.md",
"width": 580,
"height": 276
},
{
"file": "assets/readme/overview-spatial-v3.svg",
"title": "RuView sensing overview",
"target": "docs/user-guide.md",
"width": 1200,
"height": 396
},
{
"file": "assets/readme/header-features-spatial-v3.svg",
"title": "Explore the capabilities",
"target": "docs/user-guide.md",
"width": 1200,
"height": 148
},
{
"file": "assets/readme/header-edge-spatial-v3.svg",
"title": "Intelligence at the edge",
"target": "firmware/esp32-csi-node/README.md",
"width": 1200,
"height": 148
},
{
"file": "assets/readme/header-learning-spatial-v3.svg",
"title": "From recordings to models",
"target": "docs/user-guide.md",
"width": 1200,
"height": 148
},
{
"file": "assets/readme/header-evidence-spatial-v3.svg",
"title": "Evidence before accuracy claims",
"target": "docs/benchmarks/pose-estimation-cog.md",
"width": 1200,
"height": 148
},
{
"file": "assets/readme/header-docs-spatial-v3.svg",
"title": "Choose your next step",
"target": "docs/user-guide.md",
"width": 1200,
"height": 148
},
{
"file": "assets/readme/cognitum-spatial-v3.svg",
"title": "Explore Cognitum One",
"target": "https://cognitum.one",
"width": 1200,
"height": 304
}
]
}
+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">AI toolkit</title><desc id="desc">Open AI toolkit. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 0L0 -33" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="0" cy="-33" r="7" stroke="#78cfff" fill="none" class="breathe"/><path d="M0 0L-35 23" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="-35" cy="23" r="7" stroke="#78cfff" fill="none" class="breathe"/><path d="M0 0L35 23" fill="none" stroke="#438fff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="35" cy="23" r="7" stroke="#78cfff" fill="none" class="breathe"/><circle cx="0" cy="0" r="15" stroke="#78cfff" fill="none" /><path d="M-6 0L-1 5L8 -6" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="90" y="98" font-size="16" fill="#f3f7ff" text-anchor="middle" font-weight="500" >AI toolkit</text></svg>

After

Width:  |  Height:  |  Size: 3.1 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">AI toolkit</title><desc id="desc">Open AI toolkit. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 0L0 -33" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="0" cy="-33" r="7" stroke="#72eee0" fill="none" class="breathe"/><path d="M0 0L-35 23" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="-35" cy="23" r="7" stroke="#72eee0" fill="none" class="breathe"/><path d="M0 0L35 23" fill="none" stroke="#70b6ff" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" class="flow"/><circle cx="35" cy="23" r="7" stroke="#72eee0" fill="none" class="breathe"/><circle cx="0" cy="0" r="15" stroke="#72eee0" fill="none" /><path d="M-6 0L-1 5L8 -6" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="90" y="98" font-size="16" fill="#ecfafb" text-anchor="middle" font-weight="500" >AI toolkit</text></svg>

After

Width:  |  Height:  |  Size: 2.9 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Hardware</title><desc id="desc">Open Hardware. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><rect x="-25" y="-25" width="50" height="50" rx="7" fill="#0c2242" stroke="#78cfff"/><rect x="-13" y="-13" width="26" height="26" rx="3" fill="none" stroke="#438fff"/><path d="M-16 -36V-25M-16 25V36M-36 -16H-25M25 -16H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M0 -36V-25M0 25V36M-36 0H-25M25 0H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M16 -36V-25M16 25V36M-36 16H-25M25 16H36" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="0" r="4" stroke="#78cfff" fill="#78cfff" class="breathe"/></g></g><text x="90" y="98" font-size="16" fill="#f3f7ff" text-anchor="middle" font-weight="500" >Hardware</text></svg>

After

Width:  |  Height:  |  Size: 3.0 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Hardware</title><desc id="desc">Open Hardware. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><rect x="-25" y="-25" width="50" height="50" rx="7" fill="#0a2630" stroke="#72eee0"/><rect x="-13" y="-13" width="26" height="26" rx="3" fill="none" stroke="#70b6ff"/><path d="M-16 -36V-25M-16 25V36M-36 -16H-25M25 -16H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M0 -36V-25M0 25V36M-36 0H-25M25 0H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><path d="M16 -36V-25M16 25V36M-36 16H-25M25 16H36" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="0" r="4" stroke="#72eee0" fill="#72eee0" class="breathe"/></g></g><text x="90" y="98" font-size="16" fill="#ecfafb" text-anchor="middle" font-weight="500" >Hardware</text></svg>

After

Width:  |  Height:  |  Size: 2.8 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Smart home</title><desc id="desc">Open Smart home. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-37 -3L0 -34L37 -3M-28 -10V32H28V-10M-9 32V8H9V32" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-9" r="5" stroke="#438fff" fill="none" class="breathe"/></g></g><text x="90" y="98" font-size="16" fill="#f3f7ff" text-anchor="middle" font-weight="500" >Smart home</text></svg>

After

Width:  |  Height:  |  Size: 2.6 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Smart home</title><desc id="desc">Open Smart home. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-37 -3L0 -34L37 -3M-28 -10V32H28V-10M-9 32V8H9V32" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><circle cx="0" cy="-9" r="5" stroke="#70b6ff" fill="none" class="breathe"/></g></g><text x="90" y="98" font-size="16" fill="#ecfafb" text-anchor="middle" font-weight="500" >Smart home</text></svg>

After

Width:  |  Height:  |  Size: 2.4 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Models</title><desc id="desc">Open Models. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-42 6H-27L-18 -10L-9 25L3 -25L14 10L24 -3L31 6H43" fill="none" stroke="#b8f36c" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/><path d="M-42 31H43" fill="none" stroke="#233f63" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="90" y="98" font-size="16" fill="#f3f7ff" text-anchor="middle" font-weight="500" >Models</text></svg>

After

Width:  |  Height:  |  Size: 2.6 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Models</title><desc id="desc">Open Models. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-42 6H-27L-18 -10L-9 25L3 -25L14 10L24 -3L31 6H43" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/><path d="M-42 31H43" fill="none" stroke="#214451" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="90" y="98" font-size="16" fill="#ecfafb" text-anchor="middle" font-weight="500" >Models</text></svg>

After

Width:  |  Height:  |  Size: 2.4 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Cognitum</title><desc id="desc">Open Cognitum. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 -38L31 -20V18L0 38L-31 18V-20ZM-31 -20L0 -1L31 -20M0 -1V38" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><path d="M-15 1L-5 11L17 -12" fill="none" stroke="#438fff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/></g></g><text x="90" y="98" font-size="16" fill="#f3f7ff" text-anchor="middle" font-weight="500" >Cognitum</text></svg>

After

Width:  |  Height:  |  Size: 2.6 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Cognitum</title><desc id="desc">Open Cognitum. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M0 -38L31 -20V18L0 38L-31 18V-20ZM-31 -20L0 -1L31 -20M0 -1V38" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" /><path d="M-15 1L-5 11L17 -12" fill="none" stroke="#70b6ff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="trace"/></g></g><text x="90" y="98" font-size="16" fill="#ecfafb" text-anchor="middle" font-weight="500" >Cognitum</text></svg>

After

Width:  |  Height:  |  Size: 2.5 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Start here</title><desc id="desc">Open Start here. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#2d4e77"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-10.4 -5.2Q0 -18.2 10.4 -5.2" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-0.65s"/><path d="M-20.0 -10.0Q0 -35.0 20.0 -10.0" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.25s"/><path d="M-30.400000000000002 -15.200000000000001Q0 -53.199999999999996 30.400000000000002 -15.200000000000001" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.9s"/><circle cx="0" cy="20" r="4" stroke="#78cfff" fill="#78cfff"/><path d="M0 12V-1" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="90" y="98" font-size="16" fill="#f3f7ff" text-anchor="middle" font-weight="500" >Start here</text></svg>

After

Width:  |  Height:  |  Size: 3.1 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="180" height="116" viewBox="0 0 180 116" role="img" aria-labelledby="title desc"><title id="title">Start here</title><desc id="desc">Open Start here. Animated navigation icon.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="179" height="115" rx="18" fill="url(#wash)" stroke="#245360"/><g transform="translate(90 45) scale(0.65)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-10.4 -5.2Q0 -18.2 10.4 -5.2" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-0.65s"/><path d="M-20.0 -10.0Q0 -35.0 20.0 -10.0" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.25s"/><path d="M-30.400000000000002 -15.200000000000001Q0 -53.199999999999996 30.400000000000002 -15.200000000000001" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.9s"/><circle cx="0" cy="20" r="4" stroke="#72eee0" fill="#72eee0"/><path d="M0 12V-1" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="90" y="98" font-size="16" fill="#ecfafb" text-anchor="middle" font-weight="500" >Start here</text></svg>

After

Width:  |  Height:  |  Size: 2.9 KiB

+21
View File
@@ -0,0 +1,21 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="396" viewBox="0 0 1200 396" role="img" aria-labelledby="title desc"><title id="title">RuView sensing overview</title><desc id="desc">Conceptual sequence: capture RF, extract context, act locally. Links to architecture and user documentation.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#102749"/><stop offset=".55" stop-color="#071429"/><stop offset="1" stop-color="#03091a"/></linearGradient><linearGradient id="accent"><stop stop-color="#9ee7ff"/><stop offset="1" stop-color="#348bff"/></linearGradient><radialGradient id="halo"><stop stop-color="#348bff" stop-opacity=".17"/><stop offset="1" stop-color="#348bff" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
.wave{transform-box:fill-box;transform-origin:center;animation:wave 4s ease-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@keyframes wave{0%{transform:scale(.5);opacity:.7}100%{transform:scale(1.3);opacity:0}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="395" rx="18" fill="url(#wash)" stroke="#2d4e77"/><text x="36" y="44" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" class="mono">SIGNAL TO CONTEXT</text><text x="36" y="89" font-size="34" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >A sensing loop you can inspect.</text><rect x="36" y="120" width="336" height="220" rx="14" fill="#091b35" stroke="#315685"/><text x="60" y="153" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" >01</text><text x="60" y="188" font-size="26" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Capture RF</text><text x="60" y="216" font-size="15" fill="#b0c2db" text-anchor="start" font-weight="500" >CSI from supported hardware</text><g transform="translate(305 278) scale(0.35)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(305 270) scale(0.5)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-10.4 -5.2Q0 -18.2 10.4 -5.2" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-0.65s"/><path d="M-20.0 -10.0Q0 -35.0 20.0 -10.0" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.25s"/><path d="M-30.400000000000002 -15.200000000000001Q0 -53.199999999999996 30.400000000000002 -15.200000000000001" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.9s"/><circle cx="0" cy="20" r="4" stroke="#78cfff" fill="#78cfff"/><path d="M0 12V-1" fill="none" stroke="#78cfff" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="60" y="311" font-size="13" fill="#b0c2db" text-anchor="start" font-weight="500" >Provision and calibrate</text><path d="M380 229H420M412 221L420 229L412 237" fill="none" stroke="#78cfff" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="flow"/><rect x="432" y="120" width="336" height="220" rx="14" fill="#091b35" stroke="#315685"/><text x="456" y="153" font-size="12" fill="#78cfff" text-anchor="start" font-weight="500" >02</text><text x="456" y="188" font-size="26" fill="#f3f7ff" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Extract context</text><text x="456" y="216" font-size="15" fill="#b0c2db" text-anchor="start" font-weight="500" >Signal processing and models</text><g transform="translate(701 278) scale(0.35)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0b1b35" stroke="#2e5b8e"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#4473a5"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#438fff" fill-opacity=".12" stroke="#78cfff" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#438fff" fill="#438fff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#438fff" fill="#438fff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#438fff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#9ee7ff"/></g><ellipse cx="0" cy="36" rx="85" ry="28" fill="none" stroke="#438fff" class="wave"/></g><g transform="translate(701 270) scale(0.5)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#142c50" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#2e527d" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g Line truncated

After

Width:  |  Height:  |  Size: 9.7 KiB

+19
View File
@@ -0,0 +1,19 @@
<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="396" viewBox="0 0 1200 396" role="img" aria-labelledby="title desc"><title id="title">RuView sensing overview</title><desc id="desc">Conceptual sequence: capture RF, extract context, act locally. Links to architecture and user documentation.</desc><defs><linearGradient id="wash" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#0b303a"/><stop offset=".55" stop-color="#06171e"/><stop offset="1" stop-color="#030c13"/></linearGradient><linearGradient id="accent"><stop stop-color="#91f6df"/><stop offset="1" stop-color="#26bad7"/></linearGradient><radialGradient id="halo"><stop stop-color="#35d4db" stop-opacity=".17"/><stop offset="1" stop-color="#35d4db" stop-opacity="0"/></radialGradient></defs><style>
text{font-family:"Avenir Next","Segoe UI",Helvetica,Arial,sans-serif}
.mono{font-family:"SFMono-Regular",Consolas,"Liberation Mono",monospace;letter-spacing:1.6px}
.float{animation:float 5s ease-in-out infinite}
.flow{stroke-dasharray:5 12;animation:flow 5s linear infinite}
.breathe{animation:breathe 4s ease-in-out infinite}
.trace{stroke-dasharray:400;animation:trace 6s ease-in-out infinite}
.scan{animation:scan 7s ease-in-out infinite}
.orbit{transform-box:fill-box;transform-origin:center;animation:orbit 18s linear infinite}
.spark{animation:spark 3.5s ease-in-out infinite}
@keyframes float{0%,100%{transform:translate(0,0)}50%{transform:translate(3px,-10px)}}
@keyframes scan{0%,100%{transform:translateY(0)}50%{transform:translateY(-76px)}}
@keyframes orbit{to{transform:rotate(360deg)}}
@keyframes spark{50%{opacity:.25}}
@keyframes flow{to{stroke-dashoffset:-136}}
@keyframes breathe{50%{opacity:.4}}
@keyframes trace{0%,100%{stroke-dashoffset:0}50%{stroke-dashoffset:90}}
@media(prefers-reduced-motion:reduce){*{animation:none!important}}
</style><rect x=".5" y=".5" width="1199" height="395" rx="18" fill="url(#wash)" stroke="#245360"/><text x="36" y="44" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" class="mono">SIGNAL TO CONTEXT</text><text x="36" y="89" font-size="34" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >A sensing loop you can inspect.</text><rect x="36" y="120" width="336" height="220" rx="14" fill="#071b23" stroke="#285360"/><text x="60" y="153" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" >01</text><text x="60" y="188" font-size="26" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Capture RF</text><text x="60" y="216" font-size="15" fill="#9abac5" text-anchor="start" font-weight="500" >CSI from supported hardware</text><g transform="translate(305 278) scale(0.35)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(305 270) scale(0.5)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-10.4 -5.2Q0 -18.2 10.4 -5.2" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-0.65s"/><path d="M-20.0 -10.0Q0 -35.0 20.0 -10.0" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.25s"/><path d="M-30.400000000000002 -15.200000000000001Q0 -53.199999999999996 30.400000000000002 -15.200000000000001" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" class="breathe" style="animation-delay:-1.9s"/><circle cx="0" cy="20" r="4" stroke="#72eee0" fill="#72eee0"/><path d="M0 12V-1" fill="none" stroke="#72eee0" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" /></g></g><text x="60" y="311" font-size="13" fill="#9abac5" text-anchor="start" font-weight="500" >Provision and calibrate</text><path d="M380 229H420M412 221L420 229L412 237" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="flow"/><rect x="432" y="120" width="336" height="220" rx="14" fill="#071b23" stroke="#285360"/><text x="456" y="153" font-size="12" fill="#72eee0" text-anchor="start" font-weight="500" >02</text><text x="456" y="188" font-size="26" fill="#ecfafb" text-anchor="start" font-weight="650" letter-spacing="-0.6" >Extract context</text><text x="456" y="216" font-size="15" fill="#9abac5" text-anchor="start" font-weight="500" >Signal processing and models</text><g transform="translate(701 278) scale(0.35)"><path d="M-82 36L0 -6L82 36L0 78Z" fill="#0a2733" stroke="#286171"/><path d="M-82 36V-52L0 -94L82 -52V36M-82 -52L0 -10L82 -52M0 -10V78" fill="none" stroke="#366d80"/><g class="scan"><path d="M-81 32L0 -10L81 32L0 74Z" fill="#5fe3d8" fill-opacity=".12" stroke="#72eee0" stroke-opacity=".7"/></g><circle cx="-40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="-15" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="40" cy="6" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="-20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="20" cy="30" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><circle cx="0" cy="49" r="2" stroke="#70b6ff" fill="#70b6ff" class="spark"/><g class="orbit"><ellipse cx="0" cy="4" rx="102" ry="30" fill="none" stroke="#70b6ff" stroke-opacity=".4" stroke-dasharray="2 8"/><circle cx="102" cy="4" r="4" fill="#91f6df"/></g></g><g transform="translate(701 270) scale(0.5)"><ellipse cx="0" cy="48" rx="43" ry="12" fill="#133a45" opacity=".5"/><path d="M-43 48L0 27L43 48L0 69Z" fill="none" stroke="#285c68" stroke-width="1" stroke-linecap="round" stroke-linejoin="round" /><g class="float"><path d="M-42 6H-27L-18 -10L-9 25L3 -25L14 10L24 -3L31 6H43" fill="none" stroke="#72eee0" stroke-width="2" stroke-linecap="round" stroke-linejoin="roundLine truncated

After

Width:  |  Height:  |  Size: 9.3 KiB

+87
View File
@@ -0,0 +1,87 @@
<svg xmlns="http://www.w3.org/2000/svg" width="640" height="360" viewBox="0 0 960 540" role="img" aria-labelledby="title desc">
<title id="title">RuView · WiFi becomes spatial awareness.</title><desc id="desc">RuView-focused animation showing WiFi sensors, spatial perception, Ruflo coordination and an illustrative ruOS sensing workspace. Room occupancy and actions are synthetic examples, not live measurements.</desc>
<defs><radialGradient id="bg"><stop stop-color="#151518" /><stop offset="1" stop-color="#07080b" /></radialGradient><radialGradient id="aura"><stop stop-color="#edc65a" stop-opacity=".23" /><stop offset=".4" stop-color="#ba8823" stop-opacity=".07" /><stop offset="1" stop-color="#ba8823" stop-opacity="0" /></radialGradient><linearGradient id="gold" x1="0" y1="0" x2="1" y2="1"><stop stop-color="#fff4b1" /><stop offset=".5" stop-color="#ebc456" /><stop offset="1" stop-color="#96702f" /></linearGradient><linearGradient id="panel" x2="0" y2="1"><stop stop-color="#191a20" /><stop offset="1" stop-color="#0d0e13" /></linearGradient><pattern id="grid" width="36" height="36" patternUnits="userSpaceOnUse"><circle cx="18" cy="18" r=".65" fill="#d4c78e" opacity=".22" /></pattern><filter id="bloom" x="-100%" y="-100%" width="300%" height="300%"><feGaussianBlur stdDeviation="4" /></filter><clipPath id="frame"><rect width="960" height="540" rx="18" /></clipPath><clipPath id="line1"><rect x="64" y="155" width="850" height="93" /></clipPath><clipPath id="line2"><rect x="64" y="244" width="850" height="100" /></clipPath><g id="crown" fill="url(#gold)"><path d="M0 5l8 7L15 0l7 12 8-7-4 23H4z" /><rect x="4" y="31" width="22" height="3" rx="1" /></g>
<style>
text{font-family:Arial,Helvetica,sans-serif;fill:#f4f2ed}.mono{font-family:monospace}.muted{fill:#888993}.gold{fill:#e8c565}.scene{opacity:0}.end{opacity:1}.wire{fill:none;stroke:#c2a155;stroke-width:1}.all{animation-duration:18s;animation-iteration-count:infinite;animation-timing-function:cubic-bezier(.22,.65,.3,1)}
@keyframes intro{0%,17%{opacity:1;transform:translateY(0)}22%,94%{opacity:0;transform:translateY(-12px)}100%{opacity:1;transform:translateY(0)}}
@keyframes network{0%,20%{opacity:0;transform:translateY(18px) scale(.98)}25%,44%{opacity:1;transform:translateY(0) scale(1)}49%,100%{opacity:0;transform:translateY(-15px) scale(1.02)}}
@keyframes work{0%,46%{opacity:0;transform:translateY(18px)}51%,66%{opacity:1;transform:translateY(0)}71%,100%{opacity:0;transform:translateY(-14px)}}
@keyframes finale{0%,69%{opacity:0}74%,92%{opacity:1}97%,100%{opacity:0}}
.intro{animation-name:intro}.network{animation-name:network;transform-origin:480px 270px}.work{animation-name:work}.end{animation-name:finale}
@keyframes reveal1{0%,69%{transform:translateY(100px)}77%,100%{transform:translateY(0)}}@keyframes reveal2{0%,73%{transform:translateY(110px)}81%,100%{transform:translateY(0)}}.headline1{animation-name:reveal1}.headline2{animation-name:reveal2}
@keyframes breathe{50%{opacity:.6;transform:scale(1.12)}}.breathe{animation:breathe 4.5s ease-in-out infinite;transform-origin:480px 239px}
@keyframes spin{to{transform:rotate(360deg)}}.orbit{animation:spin 18s linear infinite;transform-origin:480px 239px}
@keyframes send{0%,9%{transform:translate(0,0)}14%,100%{transform:translate(-14px,-19px)}}.cursor{animation-name:send}
@keyframes type{0%,2%{clip-path:inset(0 100% 0 0)}12%,100%{clip-path:inset(0 0 0 0)}}.typing{animation-name:type;animation-timing-function:steps(26,end)}
@keyframes task{0%,49%{transform:translateY(25px);opacity:0}55%,100%{transform:translateY(0);opacity:1}}.task{animation-name:task}.task2{animation-delay:.3s}.task3{animation-delay:.6s}
@keyframes grow{0%,54%{transform:scaleX(0)}63%,100%{transform:scaleX(1)}}.grow{animation-name:grow;transform-box:fill-box;transform-origin:left}
@keyframes tick{0%,62%{opacity:0}65%,100%{opacity:1}}.tick{animation-name:tick}
@keyframes trace{0%,22%{stroke-dashoffset:1}38%,100%{stroke-dashoffset:0}}.trace{stroke-dasharray:1;animation-name:trace}
@keyframes drift{50%{transform:translateY(-10px);opacity:.8}}.dust{animation:drift 9s ease-in-out infinite}
@media(prefers-reduced-motion:reduce){*{animation:none!important}.scene{opacity:0}.end{opacity:1}.moving{display:none}}
@keyframes signal{0%,13%{transform:translate(480px,205px)}20%{transform:translate(708px,301px)}26%,43%{transform:translate(480px,239px)}49%{transform:translate(215px,155px)}55%{transform:translate(479px,155px)}61%{transform:translate(743px,155px)}72%,93%{transform:translate(782px,307px)}100%{transform:translate(480px,205px)}}
.signal{animation-name:signal}.signal-glow{animation:shimmer 3s ease-in-out infinite}@keyframes shimmer{50%{opacity:.45}}
@keyframes haloExpand{0%,12%{opacity:0;transform:scale(.5)}16%{opacity:.7}23%,100%{opacity:0;transform:scale(2.5)}}.clickring{animation-name:haloExpand;transform-origin:708px 301px}
@keyframes aurora{0%,100%{transform:translate(-30px,10px) rotate(-4deg)}50%{transform:translate(30px,-10px) rotate(4deg)}}.aurora{animation:aurora 18s ease-in-out infinite;transform-origin:480px 270px}
@keyframes dash{to{stroke-dashoffset:-80}}.stream{stroke-dasharray:2 18;animation:dash 3s linear infinite}
@keyframes count{0%,52%{opacity:0;transform:translateY(8px)}58%,100%{opacity:1;transform:translateY(0)}}.detail{animation-name:count}
@keyframes buildLine{0%,51%{stroke-dashoffset:1}64%,100%{stroke-dashoffset:0}}.buildLine{stroke-dasharray:1;animation-name:buildLine}
@keyframes underline{0%,78%{transform:scaleX(0)}85%,100%{transform:scaleX(1)}}.underline{animation-name:underline;transform-origin:66px 350px}
@keyframes dock{0%,49%{opacity:0;transform:translateY(14px)}55%,100%{opacity:1;transform:translateY(0)}}.dock{animation-name:dock}
@media(prefers-reduced-motion:reduce){.signal,.clickring{display:none}}
@keyframes vortexCW{to{transform:rotate(360deg)}}@keyframes vortexCCW{to{transform:rotate(-360deg)}}
.vortex{transform-origin:480px 270px;animation:vortexCW 36s linear infinite}.counter{transform-origin:480px 270px;animation:vortexCCW 54s linear infinite}
@keyframes current{to{stroke-dashoffset:-1200}}.current{animation:current 9s linear infinite}
@keyframes bgEnergy{0%,12%,100%{opacity:.22}22%,40%{opacity:.6}52%,65%{opacity:.4}74%,84%{opacity:.7}94%{opacity:.28}}.bgEnergy{animation:bgEnergy 18s ease-in-out infinite}
@keyframes wave{0%{opacity:0;transform:scale(.3)}20%{opacity:.35}80%{opacity:.07}100%{opacity:0;transform:scale(1.9)}}.wave{transform-origin:480px 270px;animation:wave 9s cubic-bezier(.15,.35,.5,1) infinite}
@keyframes glimmer{0%,100%{opacity:.08}50%{opacity:.72}}.glimmer{animation:glimmer 6s ease-in-out infinite}
@keyframes comet{0%{opacity:0;transform:translate(-160px,70px)}12%{opacity:.65}58%{opacity:.25}75%,100%{opacity:0;transform:translate(1060px,-300px)}}.comet{animation:comet 9s linear infinite}
@keyframes beam{0%,100%{transform:translate(-150px,0) rotate(-12deg);opacity:0}35%{opacity:.18}65%{opacity:.07;transform:translate(1050px,0) rotate(-12deg)}80%{opacity:0}}.beam{animation:beam 18s ease-in-out infinite;transform-origin:center}
@media(prefers-reduced-motion:reduce){.dynamic-bg{display:none}}
.all{animation-duration:24s}
@keyframes work{0%,46%{opacity:0;transform:translateY(18px)}50%,72%{opacity:1;transform:translateY(0)}76%,100%{opacity:0;transform:translateY(-14px)}}
@keyframes finale{0%,75%{opacity:0}79%,93%{opacity:1}97%,100%{opacity:0}}
@keyframes reveal1{0%,75%{transform:translateY(100px)}82%,100%{transform:translateY(0)}}
@keyframes reveal2{0%,78%{transform:translateY(110px)}85%,100%{transform:translateY(0)}}
@keyframes signal{0%,13%{transform:translate(480px,205px)}20%{transform:translate(708px,301px)}26%,43%{transform:translate(480px,239px)}49%{transform:translate(803px,182px)}55%{transform:translate(803px,235px)}63%,72%{transform:translate(803px,288px)}80%,93%{transform:translate(782px,307px)}100%{transform:translate(480px,205px)}}
@keyframes browserOpen{0%,48%{opacity:0;transform:translateY(20px) scale(.96)}53%,100%{opacity:1;transform:translateY(0) scale(1)}}.browserOpen{animation-name:browserOpen;transform-origin:380px 230px}
@keyframes terminalOpen{0%,53%{opacity:0;transform:translateY(14px)}58%,100%{opacity:1;transform:translateY(0)}}.terminalOpen{animation-name:terminalOpen}
@keyframes terminalText{0%,57%{opacity:0}60%,100%{opacity:1}}.terminalText{animation-name:terminalText}
@keyframes done{0%,65%{opacity:0}69%,100%{opacity:1}}.done{animation-name:done}
@keyframes pointerMove{0%,50%{transform:translate(350px,275px)}55%{transform:translate(529px,267px)}59%{transform:translate(570px,325px)}66%,100%{transform:translate(738px,346px)}}.desktopPointer{animation-name:pointerMove}
@keyframes chartGrow{0%,53%{transform:scaleY(.05)}63%,100%{transform:scaleY(1)}}.chartGrow{animation-name:chartGrow;transform-box:fill-box;transform-origin:bottom}
</style><radialGradient id="readability"><stop stop-color="#08090e" stop-opacity=".58" /><stop offset=".65" stop-color="#08090e" stop-opacity=".38" /><stop offset="1" stop-color="#08090e" stop-opacity="0" /></radialGradient><linearGradient id="beamGold"><stop stop-color="#e7bf56" stop-opacity="0" /><stop offset=".5" stop-color="#efd57f" stop-opacity=".45" /><stop offset="1" stop-color="#e7bf56" stop-opacity="0" /></linearGradient></defs><g clip-path="url(#frame)"><rect width="960" height="540" fill="url(#bg)" /><rect width="960" height="540" fill="url(#grid)" opacity=".35" /><ellipse cx="490" cy="250" rx="460" ry="320" fill="url(#aura)" />
<g><g class="aurora" fill="none" stroke="#c6a457" opacity=".11"><ellipse cx="480" cy="260" rx="470" ry="156" transform="rotate(-18 480 260)" /><ellipse cx="480" cy="260" rx="480" ry="166" transform="rotate(-18 480 260)" /><ellipse cx="480" cy="260" rx="495" ry="180" transform="rotate(-18 480 260)" /></g></g><g><g class="dynamic-bg"><g class="bgEnergy"><rect class="beam" x="0" y="-120" width="160" height="850" fill="url(#beamGold)" /><ellipse cx="480" cy="270" rx="280" ry="150" fill="none" stroke="#dcb55e" stroke-width=".9" class="wave" style="animation-delay:-0s" /><ellipse cx="480" cy="270" rx="280" ry="150" fill="none" stroke="#dcb55e" stroke-width=".9" class="wave" style="animation-delay:-3s" /><ellipse cx="480" cy="270" rx="280" ry="150" fill="none" stroke="#dcb55e" stroke-width=".9" class="wave" style="animation-delay:-6s" /><g class="vortex" fill="none"><ellipse cx="480" cy="270" rx="355" ry="160" transform="rotate(0 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="355" ry="160" transform="rotate(0 480 270)" stroke="#f7d884" stroke-width="2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-0.0s" opacity=".65" /><ellipse cx="480" cy="270" rx="392" ry="184" transform="rotate(19 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="392" ry="184" transform="rotate(19 480 270)" stroke="#f7d884" stroke-width="1.2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-1.5s" opacity=".65" /><ellipse cx="480" cy="270" rx="429" ry="208" transform="rotate(38 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="429" ry="208" transform="rotate(38 480 270)" stroke="#f7d884" stroke-width="2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-3.0s" opacity=".65" /><ellipse cx="480" cy="270" rx="466" ry="232" transform="rotate(57 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="466" ry="232" transform="rotate(57 480 270)" stroke="#f7d884" stroke-width="1.2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-4.5s" opacity=".65" /><ellipse cx="480" cy="270" rx="503" ry="256" transform="rotate(76 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="503" ry="256" transform="rotate(76 480 270)" stroke="#f7d884" stroke-width="2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-6.0s" opacity=".65" /><ellipse cx="480" cy="270" rx="540" ry="280" transform="rotate(95 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="540" ry="280" transform="rotate(95 480 270)" stroke="#f7d884" stroke-width="1.2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-7.5s" opacity=".65" /></g><g class="counter" fill="none"><ellipse cx="480" cy="270" rx="355" ry="160" transform="rotate(0 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="355" ry="160" transform="rotate(0 480 270)" stroke="#f7d884" stroke-width="2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-0.0s" opacity=".65" /><ellipse cx="480" cy="270" rx="392" ry="184" transform="rotate(-19 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="392" ry="184" transform="rotate(-19 480 270)" stroke="#f7d884" stroke-width="1.2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-1.5s" opacity=".65" /><ellipse cx="480" cy="270" rx="429" ry="208" transform="rotate(-38 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="429" ry="208" transform="rotate(-38 480 270)" stroke="#f7d884" stroke-width="2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-3.0s" opacity=".65" /><ellipse cx="480" cy="270" rx="466" ry="232" transform="rotate(-57 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="466" ry="232" transform="rotate(-57 480 270)" stroke="#f7d884" stroke-width="1.2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-4.5s" opacity=".65" /><ellipse cx="480" cy="270" rx="503" ry="256" transform="rotate(-76 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="503" ry="256" transform="rotate(-76 480 270)" stroke="#f7d884" stroke-width="2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-6.0s" opacity=".65" /><ellipse cx="480" cy="270" rx="540" ry="280" transform="rotate(-95 480 270)" stroke="#b89342" stroke-width=".6" opacity=".3" /><ellipse cx="480" cy="270" rx="540" ry="280" transform="rotate(-95 480 270)" stroke="#f7d884" stroke-width="1.2" stroke-dasharray="18 6 4 420 65 800" class="current" style="animation-delay:-7.5s" opacity=".65" /></g><circle cx="10.1" cy="303.9" r="1.3" fill="#ffe4a0Line truncated
<g class="scene intro all"><circle cx="480" cy="205" r="125" fill="url(#aura)" /><circle cx="480" cy="205" r="28" stroke="#f3d576" stroke-width="1" fill="none" /><text x="480" y="126" text-anchor="middle" font-size="32" letter-spacing="-1">Your room is already sending signals.</text>
<rect x="215" y="269" width="530" height="64" rx="12" fill="url(#panel)" stroke="#554b31" /><text x="239" y="307" font-size="17" class="mono typing all">Sense. Understand. Act.</text><rect x="689" y="282" width="38" height="38" rx="9" fill="url(#gold)" /><path d="M708 307v-13m-5 5 5-5 5 5" fill="none" stroke="#222018" stroke-width="2" stroke-linecap="round" />
<path class="cursor all" d="M731 323l0 27 7-8 8 1z" fill="#faf8e7" stroke="#151515" stroke-width="1.4" />
<text x="480" y="382" text-anchor="middle" font-size="11" letter-spacing="3" class="mono muted">WIFI SENSING. NO CAMERAS.</text><g><circle class="clickring all" cx="708" cy="301" r="25" fill="none" stroke="#fae29b" /></g><g><g fill="none" stroke="#e7c467" opacity=".65"><path d="M437 214Q480 161 523 214" /><path d="M420 205Q480 133 540 205" opacity=".4" /><path d="M404 196Q480 108 556 196" opacity=".2" /></g></g></g>
<g class="scene network all"><text x="480" y="107" text-anchor="middle" font-size="30" letter-spacing="-.7">From radio signals to useful actions.</text>
<g><path d="M480 239C548 323 639 357 735 350" class="stream" fill="none" stroke="#f1d98f" stroke-width="1.5" opacity=".45" /></g><g><path d="M480 239C420 328 311 358 225 350" class="stream" fill="none" stroke="#f1d98f" stroke-width="1.5" opacity=".45" /></g><g><path d="M480 239C520 143 640 148 714 193" class="stream" fill="none" stroke="#f1d98f" stroke-width="1.5" opacity=".45" /></g><g><path d="M480 239C440 148 322 143 247 193" class="stream" fill="none" stroke="#f1d98f" stroke-width="1.5" opacity=".45" /></g><path d="M480 239C440 148 322 143 247 193" class="wire" opacity=".2" /><path d="M480 239C440 148 322 143 247 193" class="wire trace all" pathLength="1" /><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C440 148 322 143 247 193" dur="2.4s" begin="-0.0s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C440 148 322 143 247 193" dur="2.4s" begin="-0.0s" repeatCount="indefinite" /></circle></g><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C440 148 322 143 247 193" dur="2.4s" begin="-1.2s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C440 148 322 143 247 193" dur="2.4s" begin="-1.2s" repeatCount="indefinite" /></circle></g><path d="M480 239C520 143 640 148 714 193" class="wire" opacity=".2" /><path d="M480 239C520 143 640 148 714 193" class="wire trace all" pathLength="1" /><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C520 143 640 148 714 193" dur="2.4s" begin="-0.2s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C520 143 640 148 714 193" dur="2.4s" begin="-0.2s" repeatCount="indefinite" /></circle></g><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C520 143 640 148 714 193" dur="2.4s" begin="-1.4s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C520 143 640 148 714 193" dur="2.4s" begin="-1.4s" repeatCount="indefinite" /></circle></g><path d="M480 239C420 328 311 358 225 350" class="wire" opacity=".2" /><path d="M480 239C420 328 311 358 225 350" class="wire trace all" pathLength="1" /><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C420 328 311 358 225 350" dur="2.4s" begin="-0.4s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C420 328 311 358 225 350" dur="2.4s" begin="-0.4s" repeatCount="indefinite" /></circle></g><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C420 328 311 358 225 350" dur="2.4s" begin="-1.6s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C420 328 311 358 225 350" dur="2.4s" begin="-1.6s" repeatCount="indefinite" /></circle></g><path d="M480 239C548 323 639 357 735 350" class="wire" opacity=".2" /><path d="M480 239C548 323 639 357 735 350" class="wire trace all" pathLength="1" /><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C548 323 639 357 735 350" dur="2.4s" begin="-0.6000000000000001s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C548 323 639 357 735 350" dur="2.4s" begin="-0.6000000000000001s" repeatCount="indefinite" /></circle></g><g class="moving"><circle r="7" fill="#f6cf63" opacity=".5" filter="url(#bloom)"><animateMotion path="M480 239C548 323 639 357 735 350" dur="2.4s" begin="-1.8s" repeatCount="indefinite" /></circle><circle r="2.6" fill="#ffeaa1"><animateMotion path="M480 239C548 323 639 357 735 350" dur="2.4s" begin="-1.8s" repeatCount="indefinite" /></circle></g><circle cx="480" cy="239" r="98" fill="url(#aura)" class="breathe" /><circle cx="480" cy="239" r="57" fill="#111115" stroke="#6c5830" /><circle cx="480" cy="239" r="66" fill="none" stroke="#eac55e" stroke-opacity=".6" stroke-dasharray="38 170" class="orbit" /><use href="#crown" transform="translate(465 221)" /><rect x="147" y="169" width="200" height="55" rx="10" fill="url(#panel)" stroke="#51462e" /><circle cx="164" cy="191" r="3" fill="#e8c667" /><text x="177" y="195" font-size="18">WiFi sensors</text><text x="177" y="212" font-size="8" class="mono muted" letter-spacing="1.8">CAPTURE</text><rect x="614" y="169" width="200" height="55" rx="10" fill="url(#panel)" stroke="#51462e" /><circle cx="631" cy="191" r="3" fill="#e8c667" /><text x="644" y="195" font-size="18">RuView</text><text x="644" y="212" font-size="8" class="mono muted" letter-spacing="1.8">PERCEIVE</text><rect x="125" y="326" width="200" height="55" rx="10" fill="url(#panel)" stroke="#51462e" /><circle cx="142" cy="348" r="3" fill="#e8c667" /><text x="155" y="352" font-sizeLine truncated
<g class="scene work all"><g><text x="480" y="92" text-anchor="middle" font-size="28" letter-spacing="-.6">A sensing workspace, powered by RuView.</text>
<rect x="58" y="113" width="844" height="331" rx="12" fill="#090c12" stroke="#6a5931" />
<path d="M70 144H890" stroke="#292b31" /><circle cx="76" cy="129" r="3" fill="#dcbf65" /><circle cx="87" cy="129" r="3" fill="#5b5c64" /><circle cx="98" cy="129" r="3" fill="#5b5c64" /><text x="480" y="133" text-anchor="middle" class="mono muted" font-size="9">ruOS / Desktops / ruview-workspace</text>
<rect x="59" y="145" width="125" height="287" fill="#0c0e13" />
<use href="#crown" transform="translate(72 158) scale(.4)" /><text x="91" y="170" font-size="15" font-weight="700">ruOS</text>
<rect x="67" y="209" width="107" height="25" rx="5" fill="#302819" stroke="#63512c" />
<rect x="76" y="194" width="7" height="6" rx="1" fill="none" stroke="#686b79" /><text x="91" y="201" font-size="10" class="muted">Overview</text><rect x="76" y="222" width="7" height="6" rx="1" fill="none" stroke="#e1c577" /><text x="91" y="229" font-size="10" class="gold">Desktops</text><rect x="76" y="250" width="7" height="6" rx="1" fill="none" stroke="#686b79" /><text x="91" y="257" font-size="10" class="muted">Terminal</text><rect x="76" y="278" width="7" height="6" rx="1" fill="none" stroke="#686b79" /><text x="91" y="285" font-size="10" class="muted">Swarms</text><rect x="76" y="306" width="7" height="6" rx="1" fill="none" stroke="#686b79" /><text x="91" y="313" font-size="10" class="muted">Network</text><rect x="76" y="334" width="7" height="6" rx="1" fill="none" stroke="#686b79" /><text x="91" y="341" font-size="10" class="muted">Usage</text><rect x="76" y="362" width="7" height="6" rx="1" fill="none" stroke="#686b79" /><text x="91" y="369" font-size="10" class="muted">Settings</text><text x="200" y="169" font-size="14" font-weight="700">RuView workspace</text><circle cx="364" cy="165" r="3" fill="#b9cb91" /><text x="374" y="169" font-size="9" class="muted">Running</text>
<text x="882" y="167" text-anchor="end" font-size="8" class="mono gold">SYNTHETIC DEMO</text>
<rect x="197" y="183" width="469" height="244" rx="7" fill="#14131a" stroke="#32313d" /><path d="M197 373Q370 180 666 285M197 330Q440 390 666 210" fill="none" stroke="#b49548" opacity=".18" />
<g class="browserOpen all"><rect x="209" y="195" width="308" height="163" rx="7" fill="#191d26" stroke="#5a5961" />
<rect x="209" y="195" width="308" height="24" rx="7" fill="#262a33" /><text x="221" y="211" font-size="8" class="mono muted">RuView / Room overview</text><circle cx="503" cy="207" r="2" fill="#e3c477" />
<text x="225" y="241" font-size="13" font-weight="700">Room activity</text><text x="225" y="257" font-size="9" class="muted">Synthetic scene · Presence and motion</text>
<rect x="224" y="329" width="83" height="16" rx="4" fill="#4a3d22" /><text x="233" y="340" font-size="8" class="gold">Sensor connected</text><g><path d="M229 270H494V318H229zM390 270v18m0 12v18" fill="#121921" stroke="#727781" stroke-width="1.5" /><path d="M235 278h34v15h-34zM464 280h22v24h-22z" fill="#303843" stroke="#707786" /><circle cx="285" cy="300" r="4" fill="#f0cd70" /><circle cx="285" cy="300" r="16" fill="none" stroke="#d8b353" opacity=".4" /><circle cx="326" cy="287" r="5" fill="#b7cbd0" /><path d="M326 293v12m-7-8h14m-7 8-5 8m5-8 5 8" fill="none" stroke="#b7cbd0" stroke-width="2" /><path d="M292 290Q310 273 326 287M294 306Q312 312 326 302" fill="none" stroke="#e1bd5f" stroke-dasharray="3 5" class="stream" /></g></g>
<g class="terminalOpen all"><rect x="366" y="287" width="284" height="117" rx="6" fill="#090c12" stroke="#777061" /><rect x="367" y="288" width="282" height="21" rx="5" fill="#25272f" /><text x="380" y="302" font-size="9" class="mono">Terminal</text>
<text x="380" y="324" font-size="9" class="mono gold">› Room event received</text><g class="terminalText all"><text x="380" y="343" font-size="9" class="mono muted">✓ Presence detected (demo)</text><text x="380" y="358" font-size="9" class="mono muted">✓ Rule checked</text></g><text x="380" y="384" font-size="10" class="mono gold done all">✓ Action ready for review</text></g>
<rect x="680" y="183" width="207" height="244" rx="7" fill="#12151d" stroke="#33333c" /><text x="695" y="204" font-size="12" font-weight="700">Ruflo agent team</text><path d="M694 216h178" stroke="#34323a" />
<circle cx="699" cy="235" r="5" fill="#272317" stroke="#be9f54" /><text x="713" y="239" font-size="11">Sense</text><text x="713" y="254" font-size="8" class="muted">Read room events from RuView</text><circle cx="699" cy="288" r="5" fill="#272317" stroke="#be9f54" /><text x="713" y="292" font-size="11">Evaluate</text><text x="713" y="307" font-size="8" class="muted">Check the configured rule</text><circle cx="699" cy="341" r="5" fill="#272317" stroke="#be9f54" /><text x="713" y="345" font-size="11">Act</text><text x="713" y="360" font-size="8" class="muted">Request an approved action</text><g class="done all"><rect x="693" y="381" width="181" height="29" rx="5" fill="#332b1b" stroke="#89713b" /><text x="784" y="400" text-anchor="middle" font-size="10" class="gold">✓ Room event processed</text></g>
<g class="desktopPointer all"><path d="M0 0v20l5-6 8 1z" fill="#fff2ce" stroke="#18181a" stroke-width="1" /></g>
<text x="480" y="469" text-anchor="middle" class="gold" font-size="14">Example: room activity → rule check → action review.</text></g></g><g class="scene end all"><ellipse cx="810" cy="290" rx="260" ry="240" fill="url(#aura)" />
<text x="66" y="133" class="mono gold" font-size="11" letter-spacing="3">RUVIEW PERCEPTION. RUFLO COORDINATION.</text>
<g clip-path="url(#line1)"><text class="headline1 all" x="61" y="233" font-size="72" font-weight="700" letter-spacing="-3">WiFi becomes</text></g>
<g clip-path="url(#line2)"><text class="headline2 all" x="61" y="322" font-size="72" font-weight="700" letter-spacing="-3">spatial awareness<tspan class="gold">.</tspan></text></g>
<path d="M66 382H894" stroke="#3b3528" /><text x="66" y="420" class="mono gold" font-size="17">ruos.cognitum.one</text><text x="892" y="420" text-anchor="end" font-size="10" class="mono muted" letter-spacing="2">BY COGNITUM</text><g><path class="underline all" d="M66 350h310" stroke="url(#gold)" stroke-width="2" /></g></g>
<path d="M40 490H920" stroke="#26242a" /><text x="40" y="513" class="mono muted" font-size="9" letter-spacing="1.5">RUVIEW + RUFLO + RUOS</text><circle cx="853" cy="509" r="2.5" fill="#dfbf67" /><text x="866" y="512" class="mono gold" font-size="9">READY</text><g><g class="signal all"><circle r="45" fill="url(#aura)" class="signal-glow" /><circle r="10" fill="#ebc65d" opacity=".5" filter="url(#bloom)" /><circle r="4.5" fill="#ffe5a1" /><circle r="8" fill="none" stroke="#e9c56c" opacity=".3" /></g></g></g></svg>

After

Width:  |  Height:  |  Size: 37 KiB

+25
View File
@@ -37,6 +37,31 @@ services:
# volumes: ["/path/to/models:/app/models"]
# MODELS_DIR=/app/models
- MODELS_DIR=${MODELS_DIR:-data/models}
# Auth (issue #864). The entrypoint refuses to start (exit 64) unless
# RUVIEW_API_TOKEN is set. Export it before `docker compose up`:
# export RUVIEW_API_TOKEN=$(openssl rand -hex 32)
# Bare names pass the host value through only when it is set.
- RUVIEW_API_TOKEN
# ESP32 UDP ingest (ADR-296, issue #2089). The UDP receiver defaults to
# 127.0.0.1, which cannot receive from the published 5005/udp port.
# Set ONE source guard and the entrypoint binds 0.0.0.0:
# RUVIEW_UDP_ALLOW=<node-ip-or-cidr> restrict sources (preferred)
# RUVIEW_UDP_INSECURE_LAN=true accept the spoofing risk
# Docker Desktop and OrbStack on macOS rewrite the source address to the
# bridge gateway, so a LAN-subnet allowlist drops every frame. There,
# allow the gateway instead:
# docker network inspect <project>_default \
# --format '{{(index .IPAM.Config 0).Gateway}}'
# RUVIEW_UDP_ALLOW=<gateway>/32
# The gateway admits anything that reaches the published port, so keep
# 5005/udp off untrusted networks.
- RUVIEW_UDP_ALLOW
- RUVIEW_UDP_INSECURE_LAN
- RUVIEW_UDP_BIND
# Browsing the UI from another machine also needs the host name or IP
# in SENSING_ALLOWED_HOSTS (otherwise 421), and the TCP ports above
# published beyond 127.0.0.1.
- SENSING_ALLOWED_HOSTS
security_opt:
- no-new-privileges:true
cap_drop:
+56 -3
View File
@@ -22,6 +22,10 @@
# Default is `auto`. Set CSI_SOURCE=simulated when you want
# fake data tagged as such; never set it implicitly.
# MODELS_DIR — directory to scan for .rvf model files (default: data/models)
# RUVIEW_UDP_ALLOW / RUVIEW_UDP_INSECURE_LAN — ADR-296 UDP source guard.
# Setting either makes the entrypoint bind UDP on 0.0.0.0 so
# ESP32 frames on the published 5005/udp port can arrive.
# RUVIEW_UDP_BIND — explicit UDP bind; overrides the choice above.
set -e
# ── Issue #864: fail-closed on default posture ───────────────────────────────
@@ -34,8 +38,8 @@ set -e
#
# Escape hatches (operator must opt in explicitly):
# * Set RUVIEW_API_TOKEN to a strong secret → auth enabled on /api/v1/*.
# * Set RUVIEW_ALLOW_UNAUTHENTICATED=1 → preserves the pre-fix behaviour;
# only safe on an isolated trust boundary.
# * Set RUVIEW_ALLOW_UNAUTHENTICATED=1 (or true/yes/on, any case) →
# preserves the pre-fix behaviour; only safe on an isolated trust boundary.
# * Set RUVIEW_BIND_ADDR to a loopback / private interface → unauth is fine
# when the socket isn't reachable. The auto-bind nudges toward 127.0.0.1.
#
@@ -45,7 +49,12 @@ set -e
case "${1:-}" in
cog-ha-matter|ha-matter|homecore|homecore-server) ;;
*)
if [ -z "${RUVIEW_API_TOKEN:-}" ] && [ "${RUVIEW_ALLOW_UNAUTHENTICATED:-}" != "1" ]; then
# Same spellings the sensing-server accepts for its boolean env flags.
case "${RUVIEW_ALLOW_UNAUTHENTICATED:-}" in
1|[Tt][Rr][Uu][Ee]|[Yy][Ee][Ss]|[Oo][Nn]) __allow_unauth=1 ;;
*) __allow_unauth=0 ;;
esac
if [ -z "${RUVIEW_API_TOKEN:-}" ] && [ "$__allow_unauth" != "1" ]; then
# If the operator hasn't overridden the bind, refuse outright on
# the default 0.0.0.0. If they've nailed it to loopback (or a
# specific private address they trust), let it run.
@@ -97,6 +106,50 @@ esac
# server binary so users can just pass flags:
# docker run <image> --source esp32 --tick-ms 500
if [ "${1#-}" != "$1" ] || [ -z "$1" ]; then
# ── Issue #2089: UDP CSI ingest posture inside a container ───────────────
# ADR-296 defaults the UDP receiver to 127.0.0.1. Inside a container that
# bind can never receive from a published port, so no node ever appears.
# We bind 0.0.0.0 only when the operator has configured an ADR-296 guard
# (RUVIEW_UDP_ALLOW or RUVIEW_UDP_INSECURE_LAN); without one the server
# would refuse a routable bind, so we keep loopback and say why instead.
# An explicit RUVIEW_UDP_BIND or --udp-bind always wins.
# The binary parses RUVIEW_UDP_INSECURE_LAN itself (1/yes/on/true, any
# case); here we only need to know whether it is truthy to pick the bind.
__insecure_lan=0
case "$(printf '%s' "${RUVIEW_UDP_INSECURE_LAN:-}" | tr 'A-Z' 'a-z')" in
1|true|yes|on) __insecure_lan=1 ;;
esac
__udp_bind_arg=0
__udp_guard_arg=0
for __arg in "$@"; do
case "$__arg" in
--udp-bind|--udp-bind=*) __udp_bind_arg=1 ;;
--udp-allow|--udp-allow=*|--udp-insecure-lan) __udp_guard_arg=1 ;;
esac
done
if [ -z "${RUVIEW_UDP_BIND:-}" ] && [ "$__udp_bind_arg" = 0 ]; then
if [ -n "${RUVIEW_UDP_ALLOW:-}" ] || [ "$__insecure_lan" = 1 ] \
|| [ "$__udp_guard_arg" = 1 ]; then
export RUVIEW_UDP_BIND=0.0.0.0
echo "[entrypoint] UDP CSI receiver: binding 0.0.0.0 inside the container" >&2
echo "[entrypoint] (a source guard is set: RUVIEW_UDP_ALLOW or RUVIEW_UDP_INSECURE_LAN)." >&2
else
case "${CSI_SOURCE:-auto}" in
simulated|simulate) ;;
*)
echo "[entrypoint] NOTE: UDP CSI receiver stays on 127.0.0.1 (ADR-296 default)." >&2
echo "[entrypoint] Inside a container that bind cannot receive ESP32 frames from" >&2
echo "[entrypoint] a published 5005/udp port, so no node will appear. To ingest:" >&2
echo "[entrypoint] -e RUVIEW_UDP_ALLOW=<node-ip-or-cidr> (restrict sources), or" >&2
echo "[entrypoint] -e RUVIEW_UDP_INSECURE_LAN=true (accept spoofing risk)" >&2
echo "[entrypoint] Docker Desktop/OrbStack on macOS rewrite the source to the bridge" >&2
echo "[entrypoint] gateway, so the allowlist must name <gateway>/32, not the LAN subnet." >&2
echo "[entrypoint] See https://github.com/ruvnet/RuView/issues/2089" >&2
;;
esac
fi
fi
set -- /app/sensing-server \
--source "${CSI_SOURCE:-auto}" \
--tick-ms 100 \
+8 -6
View File
@@ -6,7 +6,9 @@ Known issues and fixes from the rebase-to-upstream branch (upstream #301).
## 1. Node not appearing in /api/v1/nodes
**Symptom:** ESP32-S3 node associates with WiFi, LED blinks, but no CSI frames arrive at the server. Node missing from `/api/v1/spatial/nodes`.
**Symptom:** ESP32-S3 node associates with WiFi, LED blinks, but no CSI frames arrive at the server. Node missing from `/api/v1/nodes`.
**Check the server first:** the UDP receiver listens on `127.0.0.1` by default (ADR-296), so a natively run `sensing-server` never hears nodes on your network unless it was started with `--udp-bind 0.0.0.0 --udp-allow <node-ip-or-cidr>` (or `--udp-insecure-lan`). The startup log line for the UDP receiver says `loopback-only` when this is the problem. For Docker, see "Receiving ESP32 frames in Docker" in the [user guide](user-guide.md).
**Root cause:** After USB flash, the node enters a limping state where WiFi associates but the UDP CSI sender silently fails. The SoftAP + mDNS stack initializes but the CSI callback never fires.
@@ -145,18 +147,18 @@ ESP32 nodes still target the host on `--target-ip <host>:5005` — no firmware r
**Symptom:** `sensing-server` starts cleanly, logs `HTTP server listening on http://localhost:3000`, but loading `http://localhost:3000/` (or `/ui/index.html`) returns `404 Not Found`. Reported in #188.
**Root cause:** The default `--ui-path ../../ui` is resolved relative to the binary's *current working directory*, not the binary location. When the binary is launched from anywhere other than `crates/wifi-densepose-sensing-server/`, the relative path doesn't reach the UI assets and Axum's static file handler returns 404.
**Root cause:** The default `--ui-path ../ui` is resolved relative to the *current working directory*, not the binary location. If it doesn't exist the server also tries `./ui` and `../../ui`. Launching from `v2/` works; from anywhere else (including `v2/crates/wifi-densepose-sensing-server/`) those paths miss the repo's UI, or hit some other `ui` directory, and Axum's static file handler returns 404.
**Fix:** Pass an absolute UI path, run the binary from the crate directory, or use the Docker image (which bundles the UI under `/app/ui`).
**Fix:** Pass an absolute UI path, run the binary from `v2/`, or use the Docker image (which bundles the UI under `/app/ui`).
```bash
# Option A — absolute path (recommended for production)
sensing-server --source esp32 --udp-port 5005 --http-port 3000 \
--ws-port 3001 --ui-path /absolute/path/to/ui
# Option B — run from the crate dir (works for local dev / cargo run)
cd v2/crates/wifi-densepose-sensing-server
cargo run -- --source esp32
# Option B — run from v2/ (works for local dev / cargo run)
cd v2
cargo run -p wifi-densepose-sensing-server -- --source esp32
# Option C — Docker (no path config needed)
docker compose -f docker/docker-compose.yml up sensing-server
@@ -4,6 +4,7 @@
|-------|-------|
| **Status** | Proposed |
| **Date** | 2026-03-01 |
| **Updated** | 2026-10-01: macOS 27 also redacts SSID without Location Services; the connected link is kept with a synthetic id, and `auto`+wifi keeps UDP bound (see Amendment 1) |
| **Deciders** | ruv |
| **Codename** | **ORCA** — OS-native Radio Channel Acquisition |
| **Relates to** | ADR-013 (Feature-Level Sensing Commodity Gear), ADR-022 (Windows WiFi Enhanced Fidelity), ADR-014 (SOTA Signal Processing), ADR-018 (ESP32 Dev Implementation) |
@@ -313,3 +314,22 @@ All verification on Mac Mini (M2 Pro, macOS 26.3).
- ADR-022: Windows WiFi Enhanced Fidelity (analogous platform adapter)
- ADR-013: Feature-Level Sensing from Commodity Gear
- Issue [#56](https://github.com/ruvnet/wifi-densepose/issues/56): macOS support request
---
## Amendment 1 (2026-10-01): SSID is also redacted; keep the connected link
**Observed on macOS 27 (M-series Mac, CoreWLAN, unsigned CLI helper).** Without Location Services, `mac_wifi --scan-once` returns `"ssid":""` and `"bssid":"00:00:00:00:00:00"`, while `rssi`, `noise`, `channel` and `tx_rate` are real. The §1 assumption that "SSID + RSSI + channel" survive no longer holds. An unsigned CLI helper never calls `CLLocationManager`, so it never appears in Location Services and cannot be granted access.
Before this amendment, `resolve_bssid` abstained when both identities were blank (to avoid merging unrelated networks), so every macOS scan was dropped and `--source wifi` produced nothing, with no error.
**Change.**
1. `--scan-once` marks its single sample `"connected": true`. It is always the one connected link, so it cannot collide with another network.
2. A redacted sample marked `connected` gets a stable, locally administered synthetic BSSID derived from a reserved key plus the channel. The SSID stays empty; none is invented. Unmarked redacted lines still abstain.
3. `--source auto` that resolves to host WiFi now also binds the UDP receiver, and `wifi_task` yields while the source is a live `esp32`. ESP32 nodes that start after the 2 s boot probe still promote the source, matching the #1004 rule for the simulator.
4. A missing helper logs one clear error with the build command instead of a warning on every tick.
**Measured (2026-10-01, this change):** `--source wifi` → source `wifi:unnamed`, sensing ticks advance. `--source auto` on a quiet port → `wifi` with `udp_receiver=true`; relaying real ADR-018 frames promoted the source to `esp32`, with all 5 nodes listed.
**Still open:** real BSSID/SSID would need a signed `.app` helper with `NSLocationWhenInUseUsageDescription` that requests authorization. The helper still lives under `archive/v1` with no build step.
@@ -4,6 +4,7 @@
|-------|-------|
| **Status** | Proposed |
| **Date** | 2026-03-02 |
| **Updated** | 2026-10-04 — multistatic fusion combines phase across nodes only when every frame declares the same shared phase reference; independent ESP32 nodes and the amplitude-only server bridge fuse amplitude only and report the reason (`phase_fusion` in `/health/ready`, issue #1752). |
| **Deciders** | ruv |
| **Codename** | **RuvSense** -- RuVector-Enhanced Sensing for Multistatic Fidelity |
| **Relates to** | ADR-012 (ESP32 Mesh), ADR-014 (SOTA Signal Processing), ADR-016 (RuVector Training), ADR-017 (RuVector Signal+MAT), ADR-018 (ESP32 Implementation), ADR-024 (AETHER Embeddings), ADR-026 (Survivor Track Lifecycle), ADR-027 (MERIDIAN Generalization) |
+55 -2
View File
@@ -3,7 +3,7 @@
| Field | Value |
|-------|-------|
| **Status** | **Accepted** (MQTT track P1–P7 + P8a + P9 + P10 shipped 2026-05-23 in PR #778, 410 lib tests, witness bundle VERIFIED) / **Proposed** (Matter SDK wiring P8b deferred to v0.7.1 per §9.10) |
| **Date** | 2026-05-23 |
| **Date** | 2026-05-23 (amended 2026-10-02: P4.5 status note, §11) |
| **Deciders** | ruv |
| **Codename** | **HA-DISCO** (MQTT) + **HA-FABRIC** (Matter) + **HA-MIND** (semantic primitives) |
| **Relates to** | ADR-018 (CSI binary frame format), ADR-021 (ESP32 vitals), ADR-031 (RuView sensing-first), ADR-039 (edge vitals packet 0xC511_0002), ADR-079 (camera ground-truth), ADR-103 (cog-person-count), ADR-110 (ESP32-C6 firmware), ADR-114 (cog-quantum-vitals) |
@@ -440,7 +440,7 @@ This means: **adding a new primitive is one file change**. No MQTT schema rev, n
| **P2** | HA discovery message emitter. New module `mqtt_discovery.rs`. Emits all entity `config` topics on connect + every `--mqtt-refresh-secs`. Schema-validated against HA's published JSON schema. | pending |
| **P3** | State publication. Subscribe to internal `tokio::broadcast` channel (the one `tx.send(json)` writes to on line 3983 of `main.rs`). Translate `edge_vitals` / `sensing_update` / `pose_data` messages into per-entity state payloads. Apply rate-limit + privacy-mode filters. | pending |
| **P4** | Integration tests: dockerised mosquitto in CI (extend `.github/workflows/firmware-qemu.yml` pattern), schema-validate every emitted config against HA's `homeassistant/components/mqtt` JSON schemas (pin to a tested HA version). Add a smoke test that brings up sensing-server in `--source mock --mqtt`, subscribes with `paho-mqtt` test client, asserts on entity creation. | pending |
| **P4.5** | **Semantic inference layer (HA-MIND).** New module `semantic_inference.rs` implementing the 10 v1 primitives from §3.12. Output broadcast channel consumed by both MQTT publisher (P3) and Matter bridge (P8). Per-primitive precision/recall baselines published to `docs/integrations/semantic-primitives-metrics.md`. Unit tests per FSM + integration tests via replay of ADR-079 paired captures. | pending |
| **P4.5** | **Semantic inference layer (HA-MIND).** New module `semantic_inference.rs` implementing the 10 v1 primitives from §3.12. Output broadcast channel consumed by both MQTT publisher (P3) and Matter bridge (P8). Per-primitive precision/recall baselines published to `docs/integrations/semantic-primitives-metrics.md`. Unit tests per FSM + integration tests via replay of ADR-079 paired captures. | partial — 6 of 10 primitives on the live MQTT publisher; see §11 |
| **P5** | Docs: new `docs/integrations/home-assistant.md` with screenshots of the HA UI after auto-discovery completes, example HA dashboard YAML (Lovelace card configs), 8 starter blueprints from §3.12.2 (distress notify, wake routine, hallway dim, elderly anomaly alert, meeting lights, bathroom fan, fall-risk escalate, auto-arm security), and the raw-channel example automations: "turn on hall light when presence ON", "send notification on fall_detected event", "log HR/BR to InfluxDB". | pending |
| **P6** | Ship `--mqtt` in the next sensing-server release (target: v0.7.0). Demo end-to-end on `cognitum-v0` against a Mosquitto add-on running on a Home Assistant OS install. Update README hardware-options table with "Works with Home Assistant" badge. | pending |
| **P7** | Matter Bridge spike: build a throwaway prototype with `matter-rs` exposing one `OccupancySensor` endpoint + one `GenericSwitch` for fall. Pair against Apple Home, Google Home, and HA's Matter integration. Decision gate: if pairing works on all three, proceed to P8; if blocked, switch to chip-tool FFI and re-spike. | pending |
@@ -667,4 +667,57 @@ Empty as of 2026-05-23. New questions discovered during implementation will be f
---
## 11. Status note: P4.5 semantic states on the live publisher (2026-10-02)
Issues [#2085](https://github.com/ruvnet/RuView/issues/2085) and [#2093](https://github.com/ruvnet/RuView/issues/2093) found that discovery announced 20 entities per device while only 6 ever received state, that `presence/state` was published on every CSI frame, and that HA device ids changed on every restart. The FSMs in `crate::semantic` existed but had no caller outside their tests. This note records what is now wired and what is not.
### 11.1 What changed
- **Semantic primitives run on the live path.** `mqtt::planner` holds one `SemanticBus` per node, sampled at 1 Hz (the rate the FSM constants assume). The bus feeds only the MQTT publisher. Matter does not consume it yet, so the "output broadcast channel shared with Matter" in P4.5 is not built.
- **Announce only what has a source.** Six primitives are announced: `room_active`, `no_movement`, `elderly_inactivity_anomaly`, `someone_sleeping`, `possible_distress` and `fall_risk_elevated`. Four are not: `meeting_in_progress` needs a 1–20 % motion level, but per-node motion is a three-level classification; `bathroom_occupied`, `bed_exit` and `multi_room_transition` need zones, and the broadcast carries none (the §3.12.5 zone file is not implemented). For the same reason `zone_occupancy` (§3.1) and `pose` are not announced: `pose_keypoints` is never populated on the broadcast, and `persons[].keypoints` are synthesized.
- **Per-entity availability (§3.6).** An announced entity is `online` only while its input exists: vitals while the calibration-gated estimate is published, `fall` while the node sends `edge_vitals`, `motion_energy` and `rssi` while present, and semantic primitives after the 60 s warmup (sleeping and distress also need breathing or heart rate, fall risk needs the fall detector). Otherwise it is `offline` and no state is sent. Transitions publish at once; all availability topics are re-sent every 30 s.
- **No default values.** `motion_energy` was always `0.0` (MEASURED: 176 of 176 samples) because the bridge never set it. It now carries the node's `motion_band_power`, or the room value, or nothing. `fall` is driven by the rising edge of `edge_vitals.fall_detected`. Only `sensing_update` produces snapshots, which removes the phantom aggregate device that `edge_vitals` frames created. `present_still` now maps to 0 % motion instead of 100 %.
- **Rates (§3.7).** Binary states publish on change, plus a 60 s retained heartbeat, as §3.5 already specified. Sensors keep their configured rates.
- **Stable device identity (§3.8).** The default client id is now `wifi-densepose-<8 hex>`, persisted in `<data-dir>/mqtt_client_id`. §3.8 specified `wifi-densepose-<hostname>`; a persisted random id avoids a hostname dependency and survives hostname changes. `--mqtt-client-id` still overrides it.
### 11.2 Evidence
Unit tests (CODE-DERIVED, run with `cargo test -p wifi-densepose-sensing-server --features mqtt --lib -- mqtt:: semantic::`): `mqtt::planner::tests` covers the entity → state mapping, unavailable-without-source, warmup, on-change with heartbeat at a 73 frame/s input, sensor rates, reset and privacy mode. `mqtt::bridge::tests` covers message-type filtering, fall edges and motion energy. `mqtt::config::tests` covers client-id persistence across two "restarts" in a temp dir, override and malformed-file recovery.
Live measurement (MEASURED, 2026-10-02): one ESP32 node streaming CSI and `edge_vitals` over UDP to a macOS host; `sensing-server --source esp32 --mqtt` against `eclipse-mosquitto:2` in Docker on a non-default port; every topic captured with a `paho-mqtt` subscriber on `#`. "Before" is `03a89793` (main), "after" is this change, built `--release --features mqtt`, same command and window.
```bash
# Broker: mosquitto.conf = "listener 1883 0.0.0.0 / allow_anonymous true / persistence false"
docker run -d --rm -p 127.0.0.1:18830:1883 \
-v "$PWD/mosquitto.conf:/mosquitto/config/mosquitto.conf:ro" eclipse-mosquitto:2
# Run A (180 s, then SIGTERM) and run B (restart, 45 s) share one --data-dir.
# A paho-mqtt subscriber on '#' starts 2 s before each run.
RUVIEW_API_TOKEN=<random> sensing-server --source esp32 --udp-port 5006 \
--udp-bind 0.0.0.0 --udp-allow <node-subnet> --http-port 3000 --ws-port 3001 \
--no-mdns --data-dir <dir> --mqtt --mqtt-host 127.0.0.1 --mqtt-port 18830
```
| Measure (180 s window unless noted) | Before (`03a89793`) | After |
|---|---|---|
| HA devices announced in run A | 2 (one phantom aggregate + `-node1`) | 1 (`-node1`) |
| Entities announced per device | 20 | 15 |
| Entities that received state (node device) | 6 | 10 |
| Announced entities without state that still reported `online` | 14 | 1 (`fall`: an event, and no fall occurred) |
| `presence/state` messages (node device) | 6,814 (37.9 / s) | 4 (one per change, plus 60 s heartbeat) |
| `motion_energy` values | `0.0` in 176 of 176 | node band power, 177 samples |
| Device id, run A → run B (restart, same `--data-dir`) | `wifi-densepose-<pid A>` → `<pid B>` (4 devices on the broker after B) | same persisted id in both runs (1 device) |
| `offline` published on SIGTERM | no | no (unchanged; see 11.3) |
| Same, plus `--privacy-mode` (75 s, fresh `--data-dir`) | — | 13 entities announced; no `heart_rate`, `breathing_rate` or `pose` topic for that device |
After the change, five announced entities had no state: `heart_rate` and `breathing_rate`, because no calibration had been run; `someone_sleeping` and `possible_distress`, which depend on them; and `fall`, which is an event and no fall occurred. The first four were `offline`. `fall` was `online`, because the node sent `edge_vitals`. Before the change all 14 entities without state still reported `online`. Only one node was streaming during this measurement. #2085 used five nodes and measured about 73 `presence` messages per second per node.
### 11.3 Not done
- Semantic precision/recall on real captures. The figures in `semantic-primitives-metrics.md` have no reproducer in the repository and stay CLAIMED.
- Zone and threshold files, `--semantic*` / `--no-semantic` flags on `sensing-server`, the `reason` attribute (§3.12.4), and `semantic_events.jsonl`.
- MQTT last-will and offline-on-shutdown (§3.6). The server registers no LWT and SIGTERM does not publish `offline`, so availability stays as last published after the server stops.
- Matter consumption of the semantic states.
---
*ADR-115 is the integration story that turns RuView from "another sensing platform" into "drop-in upgrade for any HA install **and** any Matter-controller home." MQTT carries the rich, differentiated telemetry; Matter carries the standardised subset across every controller ecosystem. Numbers 111 and 112 remain reserved per the project ADR-numbering policy.*
@@ -66,10 +66,64 @@ hardware claims.
timestamps, sequence behavior, calibration, maximum size, and report rate.
- Validate OpenWrt One first, then BPI-R3 4x4, before considering MT7996/MLO.
## Amendment 1 — vendor CSI on MT7981, still unvalidated
- **Date**: 2026-09-19, recorded here 2026-10-01
- **Status**: proposed. One board model (two units). The hardware gates below are still open.
The 2026-07-18 decision said the public `mt76` tree has no supported CSI export,
and that a firmware change might be required before physical CSI exists. Both
statements need a narrower reading.
Mainline `openwrt/mt76` still has no CSI export. MediaTek's public OpenWrt feed
does: vendor id `0x0ce7`, MCU event `0xc2`, patch
`1001-wifi-mt76-mt7915-csi-implement-csi-support`. On a Wavlink WL-WN586X3
Rev A (MT7981BA + MT7976CN + MT7531AE) that patch, rebased onto the `mt76`
revision pinned by OpenWrt 24.10.8, produced complex samples without a firmware
blob swap. The `0xc2` hook is already in the MT7981 firmware that the OpenWrt
24.10.8 image carries. The
rebased patch and the two userspace fixes are in `firmware/openwrt-wn586x3/`.
What the first capture was: one record, 2.4 GHz, 20 MHz, legacy OFDM, 64
subcarriers, a single chain. CSI arrived only while a client sent frames.
The rate is client traffic, not a sensor sample clock. The capture is not in
this repository.
What it was not:
- Not a calibration, a sequence check, a timestamp check, or a repeatability run.
The label on the wire is `mediatek:physical-unvalidated`. Plain `mediatek` stays
reserved for a frame that has passed those gates.
- Not a 4x4 array. Each radio is two chains. The four antennas are a 2.4 GHz pair
and a 5 GHz pair. Two routers are two receivers. Cross-box phase is not a
shared local oscillator, so it is not a beam.
- Not the OpenWrt One or the BPI-R3. Those remain the upstream-friendly targets
in the original decision. The WN586X3 is the board this path was actually run on.
`wifi-densepose-mtk-bridge` translates `mt76-vendor` JSON, or an MtkCSIdump
datagram, into ADR-267 MTC1 and sends it to the sensing server. The server keeps
one snapshot and a bounded ring of per-subcarrier amplitude and phase per
`device_id`. It runs `mediatek-amplitude-heuristic-v0` on mean and peak amplitude,
and `activity-index-v0` on per-subcarrier fluctuation. Both are `CLAIMED`. Neither
is the ESP32 vital-sign path.
One measured property supports further work, and nothing more. Offline, on six
local dump captures from two units (2.4 GHz, HT20, chains rx0 and rx1 of transmit
stream 0), the cross-chain phase `arg(H0·conj(H1))`, summed over subcarriers,
moved 0.014-0.039 rad (p50) between adjacent frames for well-sampled
transmitters. Raw per-chain phase moved about 1.56 rad (p50), which is what
uniformly random phase gives (MEASURED; reproducer
`scripts/mediatek-csi-phase-stability.py`). That shows a stable observable inside
one radio. It does not show that the observable tracks people.
The operator guide is [`docs/mediatek-router-csi.md`](../mediatek-router-csi.md).
## Links
- [ADR-123: BFLD capture path](ADR-123-bfld-capture-path-nexmon-and-esp32.md)
- [ADR-264: RTL8720F radar wire protocol](ADR-264-rtl8720f-radar-wire-protocol.md)
- [ADR-267: MediaTek MIMO CSI wire protocol](ADR-267-mediatek-mimo-csi-wire-protocol.md)
- [MediaTek router CSI guide](../mediatek-router-csi.md)
- [upstream mt76](https://github.com/openwrt/mt76)
- [OpenWrt One](https://openwrt.org/toh/openwrt/one)
- [MediaTek OpenWrt feed](https://git01.mediatek.com/openwrt/feeds/mtk-openwrt-feeds/)
@@ -34,6 +34,81 @@ Define `MTC1` version 1 as a little-endian, self-delimiting envelope:
- Flags distinguish calibrated, saturated, time-synchronized, dropped-predecessor,
and synthetic frames. Synthetic provenance cannot be cleared by downstream code.
## Amendment 1 — PPDU types `Legacy` and `Unknown`
- **Date**: 2026-09-19
- **Status**: proposed. Demonstrated on one WL-WN586X3 capture. Not a maintainer acceptance.
- **Scope**: `PpduType` discriminants only. The 72-byte header layout, CRC,
payload ordering, flags, and caps are unchanged.
### What changed
`PpduType` gains two variants:
| Value | Name | Meaning |
|---|---|---|
| 6 | `Legacy` | Pre-HT: 802.11b CCK or 802.11a/g OFDM |
| 7 | `Unknown` | A PPDU format the producer could not classify |
Values 1-5 (`Ht`, `Vht`, `HeSu`, `HeMu`, `Eht`) are untouched.
### Why
The original set assumed every CSI report would be HT or later. The first real
MT7981 capture disproved that. A Wavlink WL-WN586X3 (MT7981B + MT7976C) running
an OpenWrt 24.10.8 image built with the patches in `firmware/openwrt-wn586x3/`
reported `rx_mode = 1` (`MT_PHY_TYPE_OFDM`) for a client associated on 2.4 GHz
channel 6 at 20 MHz. The bridge had no way to express it and dropped the frame.
That was a real data loss, not a corner case. MediaTek's CSI driver treats legacy
OFDM as a first-class mode: `mt7915_vendor_csi_tone_mask` skips only
`MT_PHY_TYPE_CCK`, and its `mode_map` is keyed by `MT_PHY_TYPE_OFDM`, `_HT`,
`_VHT` and `_HE_SU`, giving OFDM its own tone-mask group
(`1001-wifi-mt76-mt7915-csi-implement-csi-support.patch:1070-1088`). Association
and management exchanges are legacy-rate by design, so a sensing deployment will
see these continuously.
`Unknown` exists so the enum's gaps (5-7 and 12 in `mt76_phy_type`) and future
modes are carried explicitly rather than guessed at or dropped. It also covers
transports that omit the field: the MtkCSIdump UDP datagram carries no `rx_mode`,
and reporting `Unknown` there is honest where defaulting to a concrete format was
not.
### Compatibility
- **Forward, not backward.** A decoder built before this amendment rejects
discriminants 6 and 7 with `UnknownPpduType`, because the original parser
rejects unknown types by design. `TryFrom<u8>` returns `Err` rather than
panicking, so an old decoder fails safely — it just drops legacy frames, and
does so silently. **Rebuild MTC1 consumers after this codec change.**
- **Scope is narrower than "the workspace".** `wifi-densepose-hardware` defines
three independent `PpduType` enums — `csi_frame.rs` (ESP32),
`qualcomm_csi.rs`, and `mediatek_csi.rs` — and this amendment touches only the
MediaTek one. Neither of the others changes, so no other vendor bridge is
affected. The complete set of `mediatek_csi::PpduType` consumers is:
`wifi-densepose-hardware`'s own `mediatek-csi-sim` binary,
`wifi-densepose-mtk-bridge`, and `wifi-densepose-sensing-server`, which renders the
value with `{:?}` and so needed no code change. In particular the
`ppdu_type > 3` bound in `bootstrap_baseline.rs` is fed from the **ESP32**
enum via `CsiGridKey::from_frame(&Esp32Frame)`, not from MediaTek frames, so
`Legacy` and `Unknown` cannot reach it.
- Because the variants were added rather than renumbered, and Rust requires
exhaustive matches, the compiler locates any consumer that must be updated.
Nothing outside the files above needed one.
- Producers must emit 6 only for frames that genuinely are pre-HT, and 7 only
where the format is truly unclassified. Neither is a dumping ground for a
mapping the producer has not done.
- The mapping from `rx_mode` to `PpduType` is now infallible, so no MTC1 producer
may drop a frame on PPDU grounds.
- Unchanged: synthetic provenance still cannot be cleared by downstream code.
### Observed dimensions
The same capture assembled to 1x1x64. That is the client's behaviour, not a
protocol limit: the handset sent single-stream frames to a 2.4 GHz 20 MHz AP.
Expect 1x2 or 2x2 on 5 GHz, and size consumers for the chipset profile's
`max_chains` rather than for what one capture happened to show.
## Consequences
### Positive
@@ -59,3 +134,8 @@ Define `MTC1` version 1 as a little-endian, self-delimiting envelope:
- [ADR-266: MediaTek Filogic CSI platform](ADR-266-mediatek-filogic-csi-platform.md)
- [ADR-018: ESP32 binary CSI framing](ADR-018-esp32-csi-frame-protocol.md)
- [ADR-264: RTL8720F radar wire protocol](ADR-264-rtl8720f-radar-wire-protocol.md)
- Amendment 1 implementation: `v2/crates/wifi-densepose-hardware/src/mediatek_csi.rs`
(codec) and `v2/crates/wifi-densepose-mtk-bridge` (`ppdu_from_rx_mode`). The
motivating capture is a lab record and is not in this tree: it contains a
client transmitter address and raw CSI. The in-repo check is a fabricated
legacy-OFDM record in `wifi-densepose-mtk-bridge/tests/hardware_capture.rs`.
@@ -0,0 +1,218 @@
# ADR-323: RTL8721Dx (Realtek AmebaDplus) CSI Wire Protocol
- **Status**: accepted
- **Date**: 2026-09-15
- **Deciders**: RuView maintainers
- **Tags**: realtek, ameba, amebadplus, rtl8721dx, csi, protocol, rust, udp
## Context
Unlike the MediaTek/Qualcomm CSI paths (ADR-266..269), which are simulator-first
because no supported public vendor interface exports complex CSI, Realtek's
official Ameba RTOS SDK (`Ameba-AIoT/ameba-rtos`, Apache-2.0) documents and
exposes a real, callback-driven CSI API for the RTL8721Dx ("AmebaDplus")
family: `wifi_csi_config()` plus the `RTW_EVENT_CSI_DONE` event delivering a
`struct rtw_event_csi_report_info` (per-tone signed I/Q, 8-bit or 16-bit
precision, selectable tone decimation, passive or active/triggered capture).
This has been confirmed directly against SDK source
(`component/wifi/api/wifi_api_event.h`) and against the official
`example/wifi/wifi_csi` example, and validated on real PKM8721DAF-C13-F10
hardware (build+flash+boot confirmed on branch `feat/rtl8721dx-csi-bringup`).
The RTL8721Dx is a 1x1 WiFi 4 part — single TX/RX chain, not MIMO — so the
MediaTek/Qualcomm wire format (Tx/Rx dimensioned matrices, chipset chain
counts) does not fit. The vendor SDK's raw struct is also not a safe host
wire format as-is: it has no CRC, a C flexible-array-member payload, and
fields sized for firmware convenience (e.g. a wrapping `u32` microsecond
timestamp) rather than network robustness.
Per the existing repo convention (ESP32/ADR-018, MediaTek/ADR-266-267,
Qualcomm/ADR-268-269, RTL8720F radar/ADR-263-264), each vendor CSI/radar
source gets its own self-delimiting envelope and its own Rust module pair
(`wifi-densepose-hardware` for framing+codec+simulator,
`wifi-densepose-sensing-server` for the bounded JSON-safe snapshot), never a
shared "generic vendor CSI" struct that would either lose vendor-specific
fields or invite magic-space collisions. `SourceKind::Realtek` (value 2) is
already assigned to the RTL8720F FMCW radar path (ADR-263/264) and must not
be reused for CSI.
## Decision
Define `RAC1` ("Realtek Ameba CSI", version 1) as a little-endian,
self-delimiting envelope, magic `0x3143_4152` (distinct from `MTC1`
`0x3143544d`, `QCS1` 0x31535143, `RTR1` 0x31525452, and all ADR-018
`0xC511000x` siblings):
Fixed 49-byte header (explicit field-by-field serialization, no `#[repr(C)]`
padding, matching the MediaTek/Qualcomm modules' style):
| Field | Type | Notes |
|---|---|---|
| magic | u32 LE | `0x3143_4152` |
| version | u8 | 1 |
| header_len | u16 LE | 49 |
| frame_len | u32 LE | header + payload + 4-byte CRC |
| node_id | u8 | disambiguates multi-board illuminator/receiver deployments; 0 = unset |
| csi_mode | u8 | 0=passive, 1=active, 2=board-to-board (`CsiMode`) |
| sequence | u32 LE | from firmware `csi_sequence` |
| timestamp_us | u32 LE | from firmware `hw_assigned_timestamp` (hardware-clock microseconds, wraps ~71 min — consumers must not assume monotonic across a capture longer than that) |
| peer_mac | [u8; 6] | firmware `mac_addr` |
| trig_mac | [u8; 6] | firmware `trig_addr` |
| channel | u8 | operating channel |
| bandwidth | u8 | 0=20MHz, 1=40MHz |
| rx_rate | u8 | firmware `rx_rate` |
| protocol_mode | u8 | 0=OFDM,1=HT,2=VHT,3=HE (`ProtocolMode`) |
| num_sub_carrier | u16 LE | tone count in payload |
| num_bit_per_tone | u8 | 16 (8-bit I + 8-bit Q) or 32 (16-bit I + 16-bit Q) |
| decimation | u8 | tone group (1/2/4/8 — every Nth tone) |
| rssi_dbm | i8 | single RF chain (1x1 hardware) |
| rxsc | u8 | sub-20 MHz channel used |
| csi_valid | u8 | 0/1 from firmware |
| flags | u8 | bit 0 = SYNTHETIC (simulator provenance, never clearable downstream) |
| payload_len | u32 LE | bytes, == `num_sub_carrier * (num_bit_per_tone / 8)` |
Payload: `num_sub_carrier` consecutive (I, Q) tone pairs, each
`num_bit_per_tone / 16` bytes per component (1 byte/component when
`num_bit_per_tone == 16`, 2 bytes/component little-endian signed when `== 32`)
— i.e. exactly the byte layout the firmware already produces in
`csi_data[]`, just framed. CRC-32/IEEE over header+payload, final 4
little-endian bytes. One envelope per UDP datagram.
Parsers reject: unknown version, wrong `header_len`, `frame_len` exceeding
the IPv4 UDP payload cap (65,507 bytes), `num_bit_per_tone` outside
`{16, 32}`, `payload_len` inconsistent with `num_sub_carrier *
(num_bit_per_tone/8)`, trailing bytes, and bad CRC — mirroring the
MediaTek/Qualcomm parser contract exactly.
Host side: add `realtek_csi.rs` to both
`wifi-densepose-hardware` (frame + codec + simulator, mirroring
`mediatek_csi.rs`/`qualcomm_csi.rs`) and `wifi-densepose-sensing-server`
(bounded `RealtekCsiSnapshot`, mirroring `QualcommCsiSnapshot`), wired into
the same UDP listener loop and `/api/v1/csi/realtek/latest` route as the
existing vendor sources. The initial Mac compatibility port keeps Realtek CSI
as a separate bounded diagnostic stream. Fresh ESP32 frames remain the primary
sensing source. This change does not add `SourceKind::RealtekCsi` to a signed
ingest contract; that requires a separate reviewed identity-bound hardware
proof. RAC1 CRC checks corruption, not origin.
Firmware side (`example/wifi/wifi_csi` derivative, built against the
official Ameba RTOS SDK, tracked outside this repository per the existing
`CLAUDE.local.md` convention — the vendor SDK itself is never vendored into
this repo): join the board to the deployment's WiFi AP in STA mode
(passive CSI against the real AP as peer, since only one physical board is
available in this bring-up), read the AP's BSSID at runtime after
association (never hardcode a MAC into firmware source), pack each
`RTW_EVENT_CSI_DONE` report into one `RAC1` datagram, and `sendto()` it over
UDP to the host sensing-server's existing CSI port.
## Consequences
### Positive
- CSI is real, MEASURED hardware output from day one — no simulator-first
bridge phase is needed the way MediaTek/Qualcomm required, because the
vendor SDK genuinely exposes the capability and we have the board in hand.
- One board is sufficient for a first useful signal (passive CSI against the
home AP); active and board-to-board modes are additive later without a
wire-format break (`csi_mode` already reserves the values).
- Reuses the bounded UDP parser and status-route pattern without changing
existing ESP32 sensing. Unsigned RAC1 diagnostics are not authorized as
calibrated presence, person count, pose, or vital evidence.
### Negative
- 1x1 hardware means no spatial (multi-antenna) diversity from a single
board; multi-board deployment is required for the differential/localization
use cases described in the original RTL8721Dx capability analysis.
- The 71-minute microsecond timestamp wraparound (inherited from the
firmware's `u32` field) means long unattended captures need host-side
sequence-based reordering, not raw timestamp arithmetic, across a wrap.
### Neutral
- The RAC1 envelope intentionally does not carry the SDK's raw
`csi_signature` (0xABCD) marker — our own `magic`/CRC supersede it as the
framing integrity check.
## Hardware gates
- [x] Confirm SDK source exposes `RTW_EVENT_CSI_DONE` / `rtw_event_csi_report_info` (done — `component/wifi/api/wifi_api_event.h`).
- [x] Build and flash a baseline image on real PKM8721DAF-C13-F10 hardware, confirm boot over serial (done — `feat/rtl8721dx-csi-bringup`).
- [x] Flash the CSI-streaming firmware and confirm `RAC1` datagrams parse on the host from the real board (done, 2026-09-15 — see below).
- [x] Record MEASURED throughput/latency/jitter from the real running firmware (done, 2026-09-15 — see below).
- [ ] A human walk-test across the link (amplitude perturbation vs. static baseline) is deferred — requires the operator physically present; not claimed until performed.
### MEASURED: real RAC1 capture and sustained throughput (2026-09-15)
Passive CSI (`RTW_CSI_MODE_NORMAL`, `RTW_CSI_TRIG_BEACON | RTW_CSI_TRIG_DATA | RTW_CSI_TRIG_QOS_DATA`)
against the deployment AP. A single `RAC1` datagram was first
cross-checked field-by-field against the firmware's own serial log of the
same report — every byte matched: magic `0x31434152`, RSSI byte exact match,
channel/bandwidth/protocol/subcarrier-count exact match, 157-byte total frame
(49-byte header + 104-byte payload + 4-byte CRC, exactly as specified above).
`RTW_CSI_MODE_RX_RESP` (estimate CSI from ACKs to the STA's own transmissions)
never fired `RTW_EVENT_CSI_DONE` across two full test runs with confirmed
real traffic (not a hang) — this mode does not work for this hardware/AP
combination and `RTW_CSI_MODE_NORMAL` with beacon+data triggers is the
working configuration.
**An earlier note here claimed `NORMAL` mode fires only once per CSI enable
— that was wrong**, caused by a debug log line gated to print only every
50th frame (`frame_count % 50 == 1`), which was misread as "only one report
total." A follow-up capture with a UDP listener running concurrently with a
board reset measured **277 real RAC1 datagrams in 40 seconds (43,489 bytes)**,
with `csi_sequence` climbing continuously and monotonically in the firmware's
own serial log across six separate 50-frame log checkpoints in one 45s
session — CSI reporting is continuous, roughly 7-12 fps, not sparse.
### MEASURED: throughput optimization pass (2026-09-15)
Formalized as a repeatable tool: `scripts/benchmark-rtl8721dx-csi.py` (listens
on the sensing-server's CSI UDP port, reports fps/throughput/RSSI/inter-frame
interval/sequence-gap stats from real received `RAC1` datagrams).
**Baseline** (`trig_period=200` [64ms], `trig_frame_ctrl=0`,
`CSI_REPORT_BUF_NUM=1`), 30s capture:
| Metric | Value |
|---|---|
| RAC1 fps | 14.67 |
| Throughput | 2,302.7 bytes/s |
| Mean inter-frame interval | 68.3 ms |
| RSSI | mean -34.4 dBm |
`trig_period` (units of 320µs) was hypothesized as the rate-limiting
throttle — the observed 68.3ms mean interval tracks the 64ms configured
period almost exactly. Confirmed by tuning:
1. `trig_period`: 200 → 15 (the SDK doc's stated minimum recommended value,
~4.8ms)
2. `trig_frame_ctrl`: 0 → `RTW_CSI_TRIG_ACK` (second trigger source: every
link-layer ACK the AP sends for the heartbeat's own outgoing traffic, on
top of the existing beacon trigger)
3. `CSI_REPORT_BUF_NUM`: 1 → 4 (a "lack of csi buf!" drop was observed even
at the slower baseline rate; more headroom needed at a higher rate)
**Tuned**, 30s capture:
| Metric | Value | vs. baseline |
|---|---|---|
| RAC1 fps | 66.40 | **4.5x** |
| Throughput | 10,424.8 bytes/s | **4.5x** |
| Mean inter-frame interval | 14.9 ms | 4.6x lower |
| Sequence gaps | 78 / ~2036 (~3.8%) | new (0 observed at the slower baseline rate) |
| RSSI | mean -31.1 dBm | (different session/position, not a controlled variable here) |
The `trig_period` hypothesis is confirmed as the primary lever. The ~3.8%
frame-loss rate at 66 fps is a reasonable place to stop for now — it is
UDP-over-WiFi loss, not a firmware defect, and is small enough not to block
downstream sensing use; revisit only if a use case needs stronger delivery
guarantees (sequence-gap-triggered re-request, TCP fallback, etc.).
## Links
- [ADR-266](ADR-266-mediatek-filogic-csi-platform.md) / [ADR-267](ADR-267-mediatek-mimo-csi-wire-protocol.md) — sibling vendor CSI ADR pair (simulator-first precedent)
- [ADR-268](ADR-268-qualcomm-atheros-csi-platform.md) / [ADR-269](ADR-269-qualcomm-csi-wire-protocol.md) — sibling vendor CSI ADR pair
- [ADR-263](ADR-263-rtl8720f-2-4ghz-fmcw-radar-platform.md) / [ADR-264](ADR-264-rtl8720f-radar-wire-protocol.md) — existing Realtek (RTL8720F radar) path; `SourceKind::Realtek` owner
- `CLAUDE.local.md` — RTL8721Dx toolchain setup, board/COM port, first bring-up log
@@ -0,0 +1,110 @@
# ADR-361: Point the whole fleet at one transmitter, to remove selection ambiguity
**Status:** Proposed — built, deployed fleet-wide, A-B-A measured. **Hypothesis CONFIRMED, adoption NOT justified.** The pairing *fraction* rises sharply; paired observations per second fall.
**Date:** 2026-09-09
**Follows:** ADR-358 (falsified), ADR-359 (mechanism confirmed, not adopted), ADR-360 (falsified).
## The hypothesis, recorded long before it could be tested
From `docs/BACKLOG.md`, "Next: the actual pairing constraint":
> Two nodes can therefore fire in the same aligned window and still pick frames
> from *different* transmitters, which can never pair under the
> `(tx_mac, rx_seq)` key. Cheapest test and likely fix: **point the whole fleet
> at one strong transmitter with the existing ADR-060 MAC filter.**
It had never been run, and the reason was mechanical rather than scientific:
**`filter_mac` was writable only by `provision.py` over USB.** It is a 6-byte
blob, absent from the `cfg_key_t` scalar table the config API validates
against, so the test would have required nine physical visits. Exposing it over
the network (see the firmware change accompanying this ADR) reduced the whole
experiment to two config pushes.
## What was measured
The associated AP was chosen on measurement, not assumption. Five transmitters
reach all nine receivers; only one carries a usable rate:
```
transmitter n_rx total_fps mean_rssi
8c:30:66:86:a4:21 9 109.81 -66.6 <- the AP, 26x the next
96:30:66:86:a4:21 9 4.24 -65.9
08:36:c9:71:ff:52 9 0.18 -75.9
64:16:66:cb:17:07 9 0.08 -81.7
10:a5:62:24:bb:30 9 0.06 -78.2
```
A-B-A, ten 30 s windows per arm. `/api/v1/fusion` counters are cumulative since
sink start, so they are **differenced over each window** and never divided as
lifetime totals — the trap already recorded for `early_drop`/`csi_fps_samples`.
The filter applied live on all nine nodes with **no restart**, so no reboot
confounds any arm.
```
arm n paired_fraction tx/s paired/s
A1 control 10 0.5668 +/- 0.0107 73.0 41.4
B filtered 10 0.7287 +/- 0.0352 33.9 24.7
A2 control 10 0.5836 +/- 0.0120 81.5 47.6
```
**Pairing fraction +0.1535 against the pooled control (+26.7% relative),
Welch t = 13.26, df = 10.5, p < 0.0001.** Reversible: A2 returned to control,
+0.0168 against A1.
Per-node confirmation that the filter did what it claims, taken on node 3 while
eight nodes were still unfiltered:
```
node 3 (filtered) AP fps 13.56 non-AP fps 0.28 non-AP share 2.0%
node 8 (control) AP fps 12.18 non-AP fps 5.66 non-AP share 31.7%
```
The AP rate is untouched; peer traffic falls to residue.
## The hypothesis is confirmed. The fix is not.
Transmitter ambiguity **was** suppressing pairing, and removing it lifts the
fraction by a quarter. That settles a question that had been open since
2026-09-07 and had resisted three other experiments.
But the backlog called this the "likely fix", and that half does not survive.
It predicted yield would fall to ~13 fps and that "the frames that remain are
pairable". Both are true, and neither is sufficient:
- **tx/s fell 56%** (77.3 -> 33.9)
- **paired observations per second fell 44%** (44.4 -> 24.7)
A higher fraction of a much smaller pie. The quantity downstream sensing
actually consumes is paired observations per second, and that went **down**.
Fraction was the wrong figure of merit, and choosing it is what made this look
like a fix rather than a diagnosis.
## Decision
**Do not adopt. Keep the branch unmerged.** Production stays unfiltered,
`gate_mode=2`, `gate_seq_period=8`.
`filter_mac` **is** kept in the config API. It costs nothing when unset, it is
now the cheapest available probe for "is this node's pairing transmitter-
limited?", and the ADR-060 filter had been effectively unreachable for the
fleet's whole life.
## What this opens
A **partial** filter is the obvious next question and is NOT what was tested
here. Restricting to a handful of widely-heard transmitters rather than exactly
one would trade less yield for some of the gain, and there is no reason to think
the optimum is at either extreme. The five 9-receiver transmitters above are the
natural candidate set, though four of them carry almost no rate.
That is a new experiment with a different independent variable, not a variant
of this one. It should be sized on **paired observations per second**, which is
the lesson this ADR paid for.
## Honest scope
Software-only. One house, one AP, nine nodes, ~15 minutes per arm, at 05:00
local with the house quiet. The effect is far larger than the within-arm
scatter and it reversed cleanly, but it has not been repeated across times of
day, and pairing is known to move with the transmitter population (see
"Transmitter population instability", 2026-09-02).
@@ -0,0 +1,69 @@
# ADR-362: Reading the on-node log is on-demand, never streamed
**Status:** Accepted
**Date:** 2026-09-09
**Follows:** the on-node log engine (persistent 128 B-record ring + coredump-to-flash)
## Context
The firmware now keeps a persistent log on the node's `storage` volume and can
write a coredump to flash. Nothing reads either one. The log is a new concept
in this tree, so there is no existing consumer to extend and the access
pattern is genuinely open.
Two shapes were considered:
1. **Ingest** — the server polls every node on a schedule, drains new records,
and keeps its own copy that the UI reads.
2. **On-demand** — the server fetches a node's log only when a human asks for
it, and keeps nothing.
## Decision
**On-demand only. No streaming, no periodic polling, no server-side copy, no
alerting.**
The log is pulled after a board hangs or crashes, or while troubleshooting a
specific node. That is the whole access pattern.
## Why
- **The persistence already lives on the node.** The ring holds ~208 days at
128 B per 300 s, and surviving a power cycle is the entire reason the module
exists — a power cycle is how a wedged node gets recovered. A second copy on
the server duplicates a durability property that is already satisfied.
- **Airtime is this fleet's binding constraint, and it is the thing being
diagnosed.** Two nodes uplinking at 1 Mbps were measured consuming 91% of
fleet airtime. A periodic log drain across nine nodes would spend exactly the
resource whose exhaustion is the most common thing you would open the log to
investigate. An instrument must not perturb what it measures.
- **Coredumps are 64 KB.** They are only ever worth moving when something has
actually crashed.
- **The use is diagnostic, not observational.** Nobody watches nine logs. You
open one node's history when that node misbehaves.
## What this implies gets built
- A server route that proxies to the node on request — `GET
/api/v1/nodes/{id}/log`, and the same for `/coredump` — authenticating with
the OTA PSK the firmware endpoints already sit behind.
- A page under the existing Tools menu to view one node's records.
- **No** scheduler, **no** new storage, **no** retention policy on the server.
## Open question, deliberately not settled here
Whether the server decodes the 128 B records into JSON, or proxies raw bytes
and lets the decoder at the edge (browser, or `node_log_read.py`) interpret
them.
Decoding server-side means the record layout is defined in two places — C and
Rust — and can drift silently. The pilot already paid for a version of this:
a field whose sentinel the host had to learn to render as `n/a`, where a
plausible wrong value would have been worse than an obvious gap. That argues
for the raw proxy and a single definition of the format.
## Non-goals
Real-time monitoring, log streaming, fleet-wide log aggregation, and alerting
on log contents are all explicitly out of scope. If a future need genuinely
requires one of them, it should revisit this ADR rather than grow into it.
@@ -0,0 +1,93 @@
# ADR 363: Native macOS USB node onboarding
## Status
Accepted and physically qualified with one attached ESP32 S3 and one attached
ESP32 C6 through an unsigned Mac Catalyst development build. Signed
distribution remains a release gate.
## Context
RuView operators currently provision ESP32 nodes with command line tooling.
The macOS app can list server accepted node identities, but it cannot establish
that an attached serial device is RuView firmware, assign a collision free node
identity, preserve its WiFi configuration, or prove that it joined the sensing
server. A UI that merely stores a friendly name would create false success.
Generic shell access from the app would also expand authority beyond the task.
Serial logs can contain operational data, and WiFi credentials must not enter
application diagnostics or provisioning receipts.
## Decision
1. Firmware 0.8.12 provides a bounded line protocol over the physical USB serial
channel. A discovery request carries a random 128 bit nonce. The device
response binds that nonce to chip family, firmware version, current node
identity, private server target, configuration state, and a pseudonymous
64 bit device digest derived from the base MAC under a RuView domain.
2. A configuration request is accepted only for the most recently issued
nonce within 60 seconds. The nonce is single use. Node identity is limited
to 1 through 255, the target is limited to RFC 1918 IPv4, and the target port
is limited to 1 through 65535.
3. Existing WiFi configuration can be preserved. New credentials use strict
base64 framing, bounded lengths, and control character rejection. Firmware
commits NVS before acknowledging and rebooting. Credentials are never
returned, logged, or included in the receipt.
4. The macOS client uses a native serial bridge. It allowlists exact character
device prefixes, configures the port directly, and does not invoke a shell,
Python, or an external process. It rechecks the device digest immediately
before configuration to reduce port substitution risk.
5. The app suggests the lowest unused server node identity and the Mac private
LAN address. Success is shown only after the configured node appears in the
server inventory as active with evidence no older than three seconds.
6. Old firmware is reported as requiring an update. The app must not claim it
provisioned a device based only on generic RuView serial logs.
7. Automatic firmware flashing is outside this decision. Flashing requires an
exact chip and port check, a private backup, a matching image, and captured
boot evidence. Release publication remains a separate maintainer action.
8. The device digest uses the PSA Crypto SHA 256 interface on ESP IDF 5.4 and
6.0. This keeps the stable domain separated identity while avoiding the
legacy Mbed TLS hash API removed by ESP IDF 6.
9. ESP32 C6 mmWave UART defaults use GPIO 4 and GPIO 5. Startup rejects any
configured mmWave pin pair that overlaps the active console RX or TX pins.
This preserves the onboarding input path after a failed mmWave probe.
## Security and privacy consequences
The app gains Mac App Sandbox USB and serial device access, but not general file
or shell authority. The protocol does not authenticate a device with a hardware
root of trust; the digest is stable identification, not a cryptographic device
certificate. Physical USB possession is therefore the trust boundary for this
version. An authenticated fleet identity remains ADR 305 scope.
The largest failure mode is a successful NVS write followed by failure to join
the LAN. The user receives an explicit bounded verification failure instead of
a success screen, while the board retains its committed settings and can be
rescanned over USB.
## Acceptance test
Connect an ESP32 with firmware 0.8.12 to a Mac that is already receiving RuView
nodes. Use only the macOS Add Sensor workflow to preserve WiFi, assign the
lowest unused identity, and set the Mac private address. Pass when the UI
reports the expected chip and firmware, the reassigned node remains fresh for
five minutes, the stale collided identity expires, app restart preserves the
profile, no server health, send, watchdog, panic, or reboot evidence appears,
and no credential appears in app logs or the provisioning receipt.
The 2026-08-31 physical run passed on an ESP32 S3. The measured witness is in
`docs/validation/2026-08-31-macos-usb-node-onboarding.md`.
The 2026-09-11 physical run passed on an ESP32 C6 over
`cu.usbserial-3120`. The Mac app flashed the image, received the nonce bound
hello receipt, preserved node identity 3, committed the server route, and
verified fresh server evidence at minus 34 dBm. The run also exposed and fixed
the GPIO 17 console RX conflict described above.
@@ -0,0 +1,96 @@
# ADR-364: BFI capture admission and continuity
- **Status**: proposed
- **Date**: 2026-09-29
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: bfi, capture, evidence, continuity, hardware
## Context
The Archer BE550, associated iPhone and MT7927 capture host have produced
readable beamforming feedback information (BFI). Monitor-mode support alone
does not establish this capability, and readable reports do not establish
continuous sensing. Router channel width also does not guarantee an identical
feedback grid in every report.
**MEASURED:** two saved trials contain 66 and 96 unique reports. During their
approximately 30-second sender intervals, rates were 2.2 and 3.2 reports/s,
with maximum silence of 8.392 and 6.437 seconds. No reports followed in the
approximately 15-second post-traffic intervals. All 162 reports received an
independent tshark angle/control/SNR comparison. Reproducer commands, capture
hashes and limitations are in the [capture results](../../tools/bfi/OPTIMIZATION.md)
and [continuity report](../../tools/bfi_learning/CAPTURE-CONTINUITY.md).
## Decision
Admit a hardware/driver/profile combination only from its actual packet
evidence. Apply these boundaries before using its output for sensing:
1. Keep originals and receipts outside Git, with raw own-network captures on
the capture host. Record hardware, driver, channel, source hashes, capture
interval and restoration outcome. Select the authorized AP/client direction
explicitly; missing provenance cannot be inferred from an SSID.
2. Enforce container, radiotap, MAC, length, fragmentation and supported-format
checks in [bfi_decode.py](../../tools/bfi/bfi_decode.py). Validate FCS when
present and reject flagged corruption; absent FCS is a recorded limitation.
Unsupported tuples remain counted rejections. Compare every admitted angle
against an independent dissector before extending a format's support claim.
3. Deduplicate retries over the whole capture before slicing time intervals.
Preserve integer timestamps and separate feedback grids. Sequence numbers
and sounding tokens do not measure lost packets. BFI steering matrices are
not full CSI; the identity product of a square unitary matrix with its
conjugate transpose cannot serve as a changing channel feature.
4. Preserve full-observation quality metrics and add separate before-traffic,
sender-interval and after-traffic metrics. Include leading/trailing silence,
report rate, maximum gap and duration-weighted occupied one-second bins.
Sender timestamps bound attempted UDP sends, not verified over-air sounding.
5. Use an initial continuity gate of reports in at least 27 of the first 30
complete one-second sender bins and no silence longer than two seconds
anywhere in the actual sender interval, repeated in three comparable trials.
Report any partial final bin separately; never count it as a complete second.
This is a project engineering gate, not a published sensing accuracy result.
The current traffic receipt lacks a PCAP hash and AP/client binding. The
analyzer checks caller-selected file associations and hashes all inputs in its
output; this establishes content integrity, not authenticated session binding.
Stronger binding is required before automated capture admission can treat
receipt association as independently verified.
## Alternatives
Monitor-mode declarations, average report rate alone and interpolating missing
reports would each hide a different failure. Retain explicit packet validation
and interval coverage instead. ESP32/Realtek CSI remains a separate modality;
multiple unsynchronized boards do not establish a coherent BFI array.
## Consequences
### Positive
Packet validity, capture continuity and sensing accuracy have separate,
reproducible acceptance criteria. Changes in denominators cannot appear as RF
performance gains.
### Negative
Strict format and continuity gates can exclude usable but sparse observations.
New PHY formats and hardware require additional captures and independent checks.
### Neutral
The current trials fail continuity. Further motion modeling requires a new
capture result; this ADR does not authorize hardware changes or establish
identity recognition.
## Validation and links
Run `python -B -m unittest discover -s tools/bfi -v` and the focused
`test_traffic_coverage.py` suite under `tools/bfi_learning`. Synthetic tests
check parsing and interval rules; acceptance additionally requires retained
real captures and the three successful trials above.
- Extends proposed [ADR-118](ADR-118-bfld-beamforming-feedback-layer-for-detection.md)
and [ADR-123](ADR-123-bfld-capture-path-nexmon-and-esp32.md); neither is superseded.
- Preserves [ADR-120](ADR-120-bfld-privacy-class-and-hash-rotation.md)'s privacy boundary.
- Public-data admission has a separate contract in
[ADR-365](ADR-365-public-bfi-dataset-and-decoder-contract.md).
@@ -0,0 +1,121 @@
# ADR-365: Public BFI dataset and decoder contract
- **Status**: proposed
- **Date**: 2026-09-29
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: bfi, datasets, decoding, provenance, benchmarks
## Context
Public activity recordings offer a separate route to BFI model development
while local capture continuity is insufficient. They introduce different frame
formats, collection assumptions and licenses. Public HAR activity labels do
not reproduce KIT's person-identification experiment.
**MEASURED:** the pinned CSI-BFI-HAR train/validation download passed exact
size/hash checks for 120 captures plus its license card. A separate body decoder
matched 13,104 quantized angles across 14 sample bodies against tshark 4.2.2 in
two independent executions. Whole-packet admission and activity training remain
incomplete. See [results](../../tools/bfi_learning/RESULTS.md) and the
[decoder audit and reproducer](../../tools/bfi_learning/PUBLIC-BFI-DECODE.md).
## Decision
### Pin provenance and partitions
Use the [author dataset](https://huggingface.co/datasets/foysalhaque/CSI-BFI-HAR-Dataset/tree/c3013b3edc5a563a9cf6759f5d464a9d9b6f456a)
at revision `c3013b3edc5a563a9cf6759f5d464a9d9b6f456a`. The
[manifest](../../tools/bfi_learning/manifests/csi-bfi-har-c3013b3-subset.json)
pins URLs, lengths, SHA-256, labels and split membership. Preserve the pinned
dataset card's GPL-3.0 declaration and attribution separately from code licenses.
| Partition | Author directory | Domain |
| --- | --- | --- |
| Training | HAR-1/BFI/M1 | Day 1, kitchen |
| Validation | HAR-3/BFI/M1 | Day 3, classroom |
| Held out | HAR-5/BFI/M1 and M2 | Day 5, living room; two views |
Keep all views of one event in the same partition. Day and room are confounded;
M1/M2 labels do not establish heterogeneous-chipset transfer. The selected
filenames cover A–T activities and P1–P3 participants.
Default downloads exclude held-out views. The bounded
[downloader](../../tools/bfi_learning/download_public_bfi.py) verifies exact
bytes before atomic completion, rehashes resumed files, retains failed partials
and enforces transfer deadlines and byte limits. Data stays outside Git;
downloaded code is never executed by the downloader.
### Keep body decoding separate from packet admission
[public_bfi_decode.py](../../tools/bfi_learning/public_bfi_decode.py) accepts
only the audited VHT MU 3x1, 80 MHz, Ng=1, codebook-1 complete action body:
1,003 bytes containing one SNR byte, 936 angle bytes and 61 opaque MU-exclusive
bytes. It exposes quantized/radian angles and tone ordinals. Frequency indices,
exclusive-field interpretation and validated matrix reconstruction remain
separate acceptance work.
Twelve sample frames have nonzero fragment-number bits. Successful tshark
decoding and valid CRCs do not resolve their semantics. Retain live MAC guards;
quarantine these frames until an explicit, independently reviewed dataset
policy can justify admission. Require lossless pcapng conversion evidence for
integer timestamps, packet lengths, link types and all original packet bytes.
Do not silently strip headers or rewrite fragment fields to gain admission.
The natural circular representation needs 1,404 features, exceeding the current
trainer's 1,024-feature limit. Approve a bounded representation or revised limit
with allocation tests before integration; do not truncate tones implicitly.
### Admit recording quality before scoring models
Inventory every manifest file, including tiny or unusable traces. Freeze a
five-second observed-interval rule: at least 20 valid unique reports, with no
inter-report gap above one second and no interpolation. Require at least two
independent recordings per activity in each partition/view group that supply
an eligible interval. Missing A–T coverage fails benchmark admission.
Before implementing the census, also freeze interval anchoring, stride and
leading/trailing gap rules: an inter-report gap bound alone does not bound
silence at interval edges. These details remain required acceptance work.
Public files lack trusted recorder start/end manifests. Observed packet spans
cannot establish unobserved leading/trailing coverage. Report rejection reasons,
eligible recording/time coverage and accuracy together. Freeze event/file groups
before generating windows; inspect held-out quality without model selection
against held-out outcomes.
## Alternatives
Reusing the HE/SU parser without format admission, weakening live fragment
guards, or silently dropping sparse classes would change the task while hiding
the change. Keep this adapter explicit and fail incomplete benchmark coverage.
## Consequences
### Positive
Data provenance, protocol support and score coverage remain independently
auditable. New public formats cannot silently weaken live capture validation.
### Negative
Sparse files or unresolved framing may prevent a benchmark despite a successful
download. Format and representation work must precede model optimization.
### Neutral
This is offline research/reference tooling. Production integration must reuse
the Rust protocols in [ADR-291](ADR-291-public-benchmark-evaluation-harness.md)
and scorecards in [ADR-317](ADR-317-benchmark-multi-domain-scorecard.md); this ADR
does not establish a second production trainer or an integrated benchmark score.
## Validation and links
Use `test_download_public_bfi.py` and `test_public_bfi_decode.py` under
`tools/bfi_learning`, then the full packet oracle and quality census specified
in the [decode audit](../../tools/bfi_learning/PUBLIC-BFI-DECODE.md) and
[benchmark protocol](../../tools/bfi_learning/RESEARCH-NEXT.md).
- Depends on [ADR-364](ADR-364-bfi-capture-admission-and-continuity.md)'s distinction
between packet validity and coverage.
- Preserves [ADR-120](ADR-120-bfld-privacy-class-and-hash-rotation.md).
- [ADR-119](ADR-119-bfld-frame-format-and-wire-protocol.md) defines a BFLD envelope;
that envelope does not replace raw 802.11 decoding.
@@ -0,0 +1,115 @@
# ADR-366: CSI controls and frozen evaluation
- **Status**: proposed
- **Date**: 2026-09-29
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: csi, evaluation, controls, holdout, evidence
## Context
The public NTU-Fi HumanID experiment uses 14 classes and a custom directory
split of 238 training, 56 validation and 546 held-out recordings. It differs
from the author's folder mapping. Physical acquisition-session independence
is unknown; absence of exact duplicates cannot establish it.
**MEASURED:** each of six LSTM/GRU runs classified 56/56 validation recordings
correctly. A fixed-alpha ridge model using only mean amplitudes also scored
56/56. In a matched three-seed GRU diagnostic, raw/original, raw/shuffled and
demeaned/original inputs each scored 56/56 for every seed; demeaned/shuffled
scored 52/56, 55/56 and 56/56. Chronological motion is unnecessary for perfect
accuracy on this validation split. The physical source of the predictive signal
remains unresolved. [Measured results and reproduction receipts](../../tools/bfi_learning/RESULTS.md)
## Decision
### Separate diagnostics from claims of generalization
Use static and temporal controls before interpreting a sequence-model score.
Keep input files, optimizer budgets and seeds matched across raw/original,
raw/shuffled, demeaned/original and demeaned/shuffled conditions. Permute whole
feature rows within one recording, using source/seed information without labels.
Never mix partitions. Mark shuffled time coordinates as administrative.
Whole-recording demeaning is an offline diagnostic that sees the complete
recording. It cannot substantiate causal streaming inference. These nondefault
controls are restricted to the custom public identity protocol and cannot alter
next-observation BFI forecasting. Fit normalization on training inputs only.
Report recording-level accuracy, macro-F1, included recordings and seed variation
alongside validation cross entropy averaged over selected windows. Recording
predictions aggregate softmax probabilities across their selected windows;
the current loss is not computed from those recording aggregates. Treat window
accuracy as secondary because overlapping windows and repeated seeds on the same
recordings are not independent participant trials. Retain session-independence
limitations beside every custom-split result.
### Freeze selection before final evaluation
Keep the [predeclared control protocol](../../tools/bfi_learning/CONTROLS.md):
the twelve 50-epoch diagnostic runs are excluded from final-model selection.
Only the six original raw/original 200-epoch LSTM/GRU runs are eligible. Select
lowest window-averaged validation cross entropy, breaking ties by earlier
completion. Do not substitute recording-aggregated loss or change this rule in
response to a held-out outcome.
Freeze source, dataset, metadata, configuration and checkpoint hashes before
the scheduled final evaluation. Load frozen weights without retraining. The
trainer creates an exclusive one-use claim before holdout scoring and retains
it even if scoring fails; a failure requires explicit review, not an automatic
retry. This claim is scoped to the selected frozen-run directory; the controller
enforces one final evaluation across the task's candidate directories. Stop
tuning against that holdout after its result is inspected. At this decision's
evidence checkpoint, final holdout metrics remain absent.
### Preserve executable provenance
Use bounded numeric NPZ inputs/checkpoints with pickle disabled and validate ZIP
central-directory bounds before allocation. Verify hashes and restore recorded
transformations during frozen evaluation. The launcher checks returned seed,
protocol, configuration and `test: null` during tuning. Save incomplete runs
and denominator changes explicitly rather than substituting successful runs.
Current Python tools are bounded offline reference experiments. Production
benchmark integration continues through the Rust protocols and per-domain
scorecards of ADR-291/317. A fair state-of-the-art comparison needs matched data,
task, representation, partition, adaptation access and evaluation; this custom
CSI result establishes neither BFI identity performance nor paper reproduction.
## Alternatives
Validation accuracy alone cannot distinguish temporal learning from static
cues. Random window splits would leak recordings across partitions. Expanding
the eligible model set after diagnostics would change the predeclared selection
procedure. Retain the frozen recording split and candidate set.
## Consequences
### Positive
Controls expose when model complexity or temporal claims lack experimental
support. The final evaluation has a traceable selection rule and one-use record.
### Negative
A saturated validation set limits further optimization, and strict one-use
evaluation makes interrupted scoring an explicit recovery decision. New-session
evidence requires new data rather than additional fitting to this split.
### Neutral
The GRU has 24.8% fewer parameters than the tested LSTM, but measured peak GPU
allocation was higher and elapsed times do not isolate an architecture speedup.
Preserve these separate resource findings when comparing models.
## Validation and links
Run the focused `test_prepare.py`, `test_train.py`, `test_static_baseline.py`
and `test_run_batch.py` suites under `tools/bfi_learning`. Synthetic isolation
and restore tests complement the retained real-data receipts; neither establishes
new-session generalization.
- Extends [ADR-291](ADR-291-public-benchmark-evaluation-harness.md) and
[ADR-317](ADR-317-benchmark-multi-domain-scorecard.md) without superseding them.
- [Comparison contract](../../tools/bfi_learning/RESEARCH.md),
[trainer](../../tools/bfi_learning/train.py) and
[overnight frozen-selection procedure](../../tools/bfi_learning/OVERNIGHT.md).
@@ -0,0 +1,110 @@
# ADR-367: Bounded research swarm and compute ownership
- **Status**: proposed
- **Date**: 2026-09-29
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: orchestration, ruflo, codex, compute, provenance
## Context
Research, implementation and independent review can run concurrently, while
GPU training, capture hardware and final evaluation need clear ownership.
Registration in a swarm registry does not prove that a worker executed a task.
Unbounded retries, duplicate schedulers or overlapping jobs can corrupt evidence
and spend resources without improving an experiment.
**MEASURED at the September 29 checkpoint:** local GPU runs and Codex reviews
completed with receipts. Federation enrollment returned `registration paused`
and relay membership was rejected; no cross-host worker accepted a task.
Shared-memory writes were blocked by the native-WAL guard. These limitations
and the actual work locations are recorded in the
[overnight procedure](../../tools/bfi_learning/OVERNIGHT.md).
## Decision
### Assign one owner to each mutable resource
The root controller owns each capture action, GPU batch, checkpoint-selection
decision and final evaluation. Delegated workers receive bounded source/review
tasks and explicit file ownership. Worker findings are untrusted proposals for
controller review. Distinguish registration, dispatch, execution and verified
completion in receipts; only execution evidence supports a completed-work claim.
Use the existing app heartbeat as the sole recurring scheduler for this task.
Finite serial GPU loops are jobs within a wake, not independent recurring
schedulers. Resume from receipts and verified lock ownership; completed runs
are not repeated merely because a wake or supervisor restarts.
### Bound execution and audit outputs
The [GPU launcher](../../tools/bfi_learning/run_batch.py) uses Linux process
groups, an exclusive batch lock, a 900-second trainer budget and a 1,020-second
child timeout. The current experiment requires at least 4 GiB free GPU memory,
caps the trainer's PyTorch allocator fraction at 20%, and bounds logs/metrics
to 8 MiB. This allocator limit is not a whole-device memory reservation.
Verify child hashes, seed, protocol, configuration and absence of test metrics
before declaring success. Preserve unrelated GPU services; cleanup targets only
owned processes. Audit matrix and batch ownership before changing stale locks.
The task-specific [Codex review wrapper](../../tools/bfi_learning/codex_iteration.py)
uses bounded source snapshots on stdin, read-only sandboxing, execution rules,
an exclusive lock, timeout and private event logs. This differs from the generic
contributor adapter's configuration contract. Check events for unexpected tool
execution; never execute suggestions automatically or grant child text authority.
Preserve federation and memory safety guards when a backend is unavailable.
Use private local receipts as the fallback and report degraded coordination.
Retry only after a relevant service/writer change or a concrete causal fix.
### Bind compute to an explicit task budget and end time
Use local capacity first. A paid GPU needs a measured capacity requirement and
an explicitly authorized cumulative budget covering actual spend, outstanding
reservations and fees. Before rental, check current prices, active instances
and remaining allowance; reserve maximum bounded cost, enforce termination and
verify resource cleanup. Transfer only authorized public licensed data and
reviewed source. A future run needs its own applicable authorization.
For the current run, the authorized cap is USD 200, local RTX 5080 capacity has
been sufficient and the recorded spend is USD 0. The existing schedule ends
September 30, 2026 at 07:00 America/Toronto. Its final wake performs the frozen
evaluation, reconciles owned jobs/resources and the ledger, records results,
then pauses the heartbeat. These are run-specific values, not durable spending
or scheduling authority granted by this ADR.
## Alternatives
Multiple controllers per GPU, detached indefinite loops, and treating a
registration response as an executed agent task make ownership unverifiable.
Retain explicit owners, finite jobs and execution receipts instead.
## Consequences
### Positive
Parallel review remains useful while resource mutation and final evaluation
have one accountable owner. Receipts support recovery without silent reruns.
### Negative
Serial GPU ownership can leave capacity unused. Federation or memory outages
reduce coordination features, and bounded jobs may need reviewed continuation.
### Neutral
Source and sanitized documentation belong in the isolated worktree. Raw RF
data, subject recordings, weights, logs, credentials and transcripts stay outside
Git. A completed experiment does not authorize model promotion, firmware
flashing, merging or publication.
## Validation and links
Run `test_run_batch.py` and `test_codex_iteration.py` under `tools/bfi_learning`.
Audit real execution/cleanup receipts, lock owners, ledger and scheduler state
at completion; successful synthetic tests alone cannot establish these outcomes.
- Preserves [ADR-283](ADR-283-ruview-community-metaharness-flywheel.md)'s
proposal-only learning and maintainer promotion boundary.
- Follows [ADR-321](ADR-321-decision-policy-action-authorization.md).
- Depends on [ADR-366](ADR-366-csi-controls-and-frozen-evaluation.md) for frozen
candidate selection and final holdout handling.
@@ -0,0 +1,187 @@
# ADR-368: RuView compute kernel as a WASM + napi-rs npm package
- **Status**: proposed
- **Date**: 2026-09-30
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: npm, wasm, napi-rs, metaharness, vitals, supply-chain
- **Related**: ADR-021 (vitals), ADR-263/265 (npm harness review and
distribution), ADR-283 (community metaharness), ADR-285 (WASM-first Homecore
metaharness)
## Context
`@ruvnet/ruview` (ADR-283) is a dependency-free operator/contributor harness:
it guides, verifies, and delegates, but it cannot run RuView signal processing.
JavaScript consumers (MCP hosts, dashboards, browser demos, agents) that want
real RuView computation must today start the Rust sensing server or re-implement
DSP in JS, which drifts from the Rust source of truth.
We want one Rust implementation reachable from npm with:
1. a portable default that works on every Node platform and in browsers;
2. an optional native path for hosts that want it;
3. the metaharness trust posture: least authority by default, the backend that
actually ran is always reported, fail-closed on missing or tampered
artifacts, and no accuracy claim without an evidence tag.
## Decision
### One core, one ABI, two transports
- **`v2/crates/ruview-kernel`** (workspace member) wraps the ADR-021 pipeline
from `wifi-densepose-vitals` (EMA preprocessing, breathing 0.1–0.5 Hz, heart
0.8–2.0 Hz, anomaly z-score) in a streaming `Session`, a deterministic
SplitMix64 synthetic-CSI generator, and a bounded session registry.
- The entire ABI is one function: `call(op, request_json) -> envelope_json`,
returning `{"ok":true,"result":…}` or
`{"ok":false,"error":{"code","message"}}`. Operations: `info`,
`validate_config`, `analyze`, `synthesize`, `session_open|push|summary|close`.
Keeping the ABI to a single string function makes backend parity a property
of the transport rather than of duplicated bindings.
- **WASM transport** (`src/wasm_abi.rs`, `wasm32-unknown-unknown` only):
`rvk_alloc`/`rvk_free`/`rvk_call`/`rvk_abi_version`. The module has **zero
imports** — no filesystem, network, clock, or randomness. The JS loader
refuses any module that imports host functions.
- **napi-rs transport** (`v2/crates/ruview-kernel-napi`): exports only
`abiVersion()` and `call(op, json)`. It is a standalone Cargo workspace
(listed in the v2 `exclude`) so napi never enters
`cargo test --workspace`, and it builds with `panic = "unwind"` plus
`catch_unwind` so a defect becomes an error envelope, not a Node crash.
### Input is untrusted at the ABI
Requests are bounded before processing: 16 MiB per request, 512 subcarriers,
100 000 frames per call or synthesis, 64 concurrent sessions. Unknown JSON
fields are rejected (`deny_unknown_fields`); frame widths must match the
configured subcarrier count; values must be finite with magnitude ≤ 1e9; all
frames of a batch are validated before any is processed. The core returns an
error envelope instead of panicking on caller data (release builds of the WASM
module use `panic = "abort"`, which the loader surfaces as a `trap` and
recovers from with a fresh instance).
### npm package `@ruvnet/ruview-kernel` (`harness/ruview-kernel`)
- Zero runtime dependencies. `loadKernel({backend})` with
`backend ∈ {wasm (default), napi, auto}` or `RUVIEW_KERNEL_BACKEND`.
WASM is the default because it carries no host authority. `napi` never
silently falls back; `auto` prefers napi and records `fallbackReason`.
- Artifacts are built from source by `npm run build` (`scripts/build.mjs`,
no shell) into `wasm/` and `native/ruview_kernel.<triple>.node`, with
`SHA256SUMS`. The loader verifies checksums; a mismatch is always fatal and
never an `auto` fallback trigger; `requireIntegrity` also refuses
unchecksummed artifacts. Built artifacts are git-ignored and never committed.
- `src/wasm.js` is filesystem-free, so browsers can use
`loadWasmKernelFromBytes(bytes)`.
- CLI `ruview-kernel`: `doctor`, `info`, `selftest`, `parity`, `synth`,
`analyze --file`, `bench`. `synth --out` refuses to overwrite files;
`analyze` reads only a bounded regular file.
- `selfTest()` runs synthesize → analyze → chunked streaming session → hostile
input, and asserts known-tone recovery. `parity()` diffs every output of
both backends with a 1e-9 relative tolerance, because `sin`/`exp` come from
different libm implementations on wasm32 and native targets; exact JSON
equality is reported separately and is not required.
### Metaharness integration
`@ruvnet/ruview` gains the read-only tool `ruview_kernel_selftest`
(CLI: `ruview kernel [--backend …]`). It dynamically imports the fixed
specifier `@ruvnet/ruview-kernel`, so `@ruvnet/ruview` keeps zero runtime
dependencies, and it returns `kernel_not_installed` when the package is absent.
MCP arguments are limited to `backend` and `seconds`; they cannot choose a
module or artifact path. The capability appears in `ruview_guidance` as
`portable-compute-kernel`.
## Evidence
- **SYNTHETIC** (`cd harness/ruview-kernel && node bin/cli.js selftest`):
60 s at 20 Hz, 56 subcarriers, 15 bpm breathing and 72 bpm heart tones →
recovered 15.0 bpm respiratory (valid) and 70.6 bpm heart (degraded
confidence 0.44). Tolerances asserted: ±3 bpm respiratory, ±8 bpm heart.
- **SYNTHETIC** parity (`node bin/cli.js parity`): 900 frames / 45 readings,
no difference above 1e-9 relative between wasm32 and linux-x64-gnu napi;
outputs are not byte-identical.
- **MEASURED** on one Linux x64 development container
(`node bin/cli.js bench --backend wasm|napi --seconds 120`): about 78 ms
(wasm) and 63 ms (napi) per 2 400-frame `analyze`, including JSON transfer.
Single-host timing, not a cross-platform claim.
- The WASM release artifact is about 230 KB and imports nothing
(`node scripts/verify-artifacts.mjs --require-wasm`).
None of these numbers says anything about real-world vital-sign accuracy.
That still requires the ADR-293 ground-truth protocol with a reference device.
## Amendment 1 (2026-10-01): binary transport and optimization evidence
**Profile (MEASURED on one Linux x64 development container, 2 400 frames ×
56 subcarriers).** JSON was the dominant cost:
- JS `JSON.stringify` of the request alone took 52–78 ms against
107–124 ms for a whole `analyze`.
- Rust then re-parsed about 270 000 numbers.
**Decision.** Add a binary fast path without changing the ABI version.
- New operations `analyze_flat` and `session_push_flat`, carried by
`call_f64(op, json, &[f64])`:
- WASM: `rvk_call_f64`, which reads bytes so alignment does not matter;
- napi: `callF64(op, json, Float64Array)`.
- `analyze()` and `session.push()` pack uniform frames into one
`Float64Array` (all amplitudes, then all phases) and use the binary path
automatically. Irregular input still takes the JSON path so its error names
the bad frame.
- `analyzeFlat(Float64Array, …)` is the zero-copy entry.
- `Session::step` is shared by both paths.
**Evidence.**
- **Byte-identical outputs.** On identical `f64` inputs the two paths
produce byte-identical JSON (`tests/abi.rs`).
- **A JSON precision finding.** serde_json's default float parser is not
correctly rounded, so the JSON transport perturbs inputs by ULPs (34 of 160
amplitude values in a 2 s, 4-subcarrier SYNTHETIC sample). The binary path is bit-exact.
- **Speed.** MEASURED, `ruview-kernel bench --seconds 120`, same host:
| Backend | JSON transport | Binary transport | Speed-up |
|---|---|---|---|
| wasm | 134.8 ms | 47.8 ms (~50 000 frames/s) | 2.82× |
| napi | 109.7 ms | 44.8 ms (~53 600 frames/s) | 2.45× |
- **Remaining cost is DSP.** Native timing gives heart-rate extraction
20.8–22.9 ms of 30.6–38.3 ms (about 60–68%, MEASURED; reproducer:
`cargo run --release -p ruview-kernel --example profile_stages`): the per-frame autocorrelation over the 15 s window
in `wifi-densepose-vitals`. That crate is shared with the production sensing
server, and an incremental ACF would change floating-point summation order
and therefore outputs. It is left as a follow-up that needs its own review
and regression evidence.
**Rejected after measurement.** MEASURED, three runs on the same host;
reproducer: `node bench/wasm-variants.mjs` in `harness/ruview-kernel`. Each
run is the median of 15 `analyzeFlat` calls on 2 400 frames. All variants
produced byte-identical outputs.
| Variant | Size | Median ms (3 runs) | Decision |
|---|---|---|---|
| baseline (`opt-level = 3`) | 244 KB | 38.3–40.6 | kept |
| `+simd128` | 242 KB | 37.4–54.7 | rejected: no consistent gain, and it needs SIMD-capable engines |
| `opt-level = "s"` | 227 KB | 40.8–70.1 | rejected: consistently slower for a 17 KB size saving |
## Consequences
- JavaScript consumers get real RuView DSP without a server, in Node or the
browser, with one Rust source of truth.
- A second native build matrix exists. Until per-platform packages
(`@ruvnet/ruview-kernel-<triple>`) are published by CI, only WASM is
portable; the loader already resolves those package names.
- JSON transfer dominates the cost of small calls; a typed-array fast path can
be added later behind a new ABI version.
## Rollout and rollback
1. CI (`ruview-kernel-npm.yml`) builds both transports from source, runs the
Rust tests, the package tests with `RUVIEW_KERNEL_REQUIRE_NAPI=1`, parity,
artifact verification, and `npm pack --dry-run`.
2. Publication stays CI-only with provenance (ADR-265) and requires explicit
maintainer approval; this ADR does not add a publish job.
3. Rollback: remove the `ruview_kernel_selftest` tool (the harness has no
dependency on the package) and deprecate the npm version; the Rust crates
are additive.
@@ -0,0 +1,74 @@
# ADR-369: npm RuView operator surface — CLI, MCP, and SDK over one registry
- **Status**: proposed
- **Date**: 2026-09-30
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: npm, mcp, sdk, cli, metaharness, least-authority
- **Related**: ADR-263/265 (npm harness review and distribution), ADR-283
(community metaharness), ADR-368 (compute kernel), ADR-370 (flashing),
ADR-371 (training), ADR-372 (doctor)
## Context
`@ruvnet/ruview` 0.5.x could onboard, lint claims, verify the proof, monitor a
node, and read Cognitum Spaces. Three operator jobs still left the package:
- **Flashing** was a Windows-only stub that returned `manual_step_required`.
- **Training** had only a skill document; nothing ran a trainer or checked
results mechanically.
- **Diagnostics**: `doctor` printed six PASS/FAIL lines without remedies.
There was also no programmatic API. Integrators had to shell out to the CLI or
speak MCP.
## Decision
1. **One registry, three front ends.** Every capability is a `ruview_*` tool
in `src/tools.js`, with a JSON schema, a `TOOL_POLICY` class, and a
structured, fail-closed result. The CLI verbs, the MCP server, and the new
SDK all go through `runTool`. There is one argument validator and one
authorization path.
2. **SDK** (`@ruvnet/ruview/sdk`, typed by `src/sdk.d.ts`). `createRuView()`
returns a frozen object with `doctor`, `firmware.{ports,plan,flash,monitor}`,
`training.{plan,run,gate,calibrate}`, `kernel.{selfTest,load}`, `guidance`,
`claimCheck`, `verify`, `spaces`, `memorySearch`, `call`, `tools`, and
`startMcpServer`.
- Calls run with `source: 'sdk'`: trusted in-process code, like the CLI.
Mutating tools still refuse to act without `confirm: true`.
- `{ strict: true }` turns `ok:false` into a `RuViewError`.
3. **Plan/act split for every mutation.**
- `ruview_firmware_plan` and `ruview_train_plan` are read-only and their
schemas do not accept `confirm`. An MCP client without write grants can
review the exact command.
- The acting tools (`ruview_node_flash`, `ruview_train`) keep the existing
grant-plus-confirm gate: `hardware-write` and `workspace-write`.
4. **New tools**: `ruview_doctor`, `ruview_firmware_plan`,
`ruview_firmware_ports`, `ruview_train`, `ruview_train_plan`,
`ruview_train_gate`. `ruview_node_flash` is re-implemented (ADR-370).
`TOOL_POLICY`, `.harness/claims.json`, and `.harness/mcp-policy.json` are
kept in lock-step, and a test fails on any drift.
5. **Process execution** for operator tools goes through `src/exec.js`:
- argv only, never a shell;
- a scrubbed environment: the base allowlist plus toolchain locations such
as `LIBTORCH`, `CARGO_HOME`, `VIRTUAL_ENV` and `IDF_PATH`, and no
credentials;
- bounded output, and secrets redacted.
6. **Zero runtime dependencies stays.** The optional compute kernel
(ADR-368) is loaded dynamically. Version 0.6.0. The unpacked-size ceiling
rises from 160 KiB to 224 KiB for the new modules, with the budget recorded
in both npm workflows.
## Consequences
- Agents, scripts, and humans get the same behaviour and the same refusals.
- The package now drives real hardware and training. Keeping the plan/act
split and argv-only execution is essential, and tests pin both.
- The surface is larger (16 MCP tools). Hosts that cap tool counts can
filter by annotations (`readOnlyHint`).
## Acceptance
- `cd harness/ruview && npm test && npm run test:security` (drift test in
`test/sdk.test.mjs`).
- `npx @ruvnet/ruview tools` lists 16 tools. `node -e "import('@ruvnet/ruview/sdk')"`
resolves from an installed tarball.
@@ -0,0 +1,83 @@
# ADR-370: Cross-platform, verified ESP32 firmware flashing from npm
- **Status**: proposed
- **Date**: 2026-09-30
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: firmware, esp32, esptool, supply-chain, hardware-evidence
- **Related**: ADR-028 (capability audit / witness), ADR-369, firmware
`README.md` §0 and §2
## Context
The documented flash procedure is four `esptool write_flash` offsets. It is
typed by hand, and a wrong chip, wrong flash size, or tampered image can brick
a board or silently run the wrong firmware. The previous `ruview_node_flash`
was Windows-only and never flashed. Project rules also say that a successful
build or write is **not** hardware validation; that requires a captured
boot/runtime log from real silicon.
## Decision
`src/firmware.js` implements plan → verify → preflight → write → evidence:
1. **Variants**:
- `s3-8mb` → `esp32s3` / 8MB
- `s3-4mb` → `esp32s3` / 4MB; prefers `*-4mb.bin` images
- `c6` → `esp32c6` / 4MB
Offsets follow the shipped partition tables: bootloader `0x0`, partition
table `0x8000`, optional OTA data `0xf000`, app `0x20000`. NVS is never
written, so WiFi and node configuration survive.
2. **Bundle verification**. A bundle is an extracted release flash bundle,
`release_bins/<variant>`, or an ESP-IDF `build/` directory.
- Every image must match `SHA256SUMS` or `SHA256SUMS.txt`. A mismatch is
always fatal.
- Missing checksums are refused unless `allow_unverified` is set, which is
for local builds only.
- Images must be regular, non-empty files of at most 16 MiB, inside the
bundle; symlinks that escape the bundle are refused.
- An 8 MB variant refuses a 4 MB partition table.
- A bundle without OTA data produces an explicit warning (a node on `ota_1`
keeps booting `ota_1`).
3. **Argv, not shell.**
- The port must match `COM<n>` or `/dev/(tty|cu)*`.
- Baud comes from a fixed set.
- The command is `python -m esptool --chip … --port … --baud … write_flash
--flash_mode dio --flash_size … <offset> <file>…`. The underscore forms
work in both esptool 4.x and 5.x.
4. **Chip preflight.** `esptool chip_id` must report the variant's chip. The
parser covers the 4.x "Chip is" and 5.x "Chip type:" / "Detecting chip
type…" formats. A mismatch or missing response stops the flow before any
write.
5. **Evidence.** After a write, a read-only pyserial capture (default 15 s,
port and duration passed as argv) is summarized: line count, CSI
callbacks, MGMT+DATA upgrade, panic markers, firmware version.
- `hardwareValidated: true` only when CSI callbacks appear with no panic;
the result's `evidence` is then `MEASURED` with a boot-log reason.
- Otherwise the result states the board is not validated.
6. **Authority.** Without `confirm` the result is a plan. MCP needs the
`hardware-write` grant plus `confirm`. `ruview_firmware_plan` and
`ruview_firmware_ports` are read-only.
7. **Provisioning stays out.** WiFi credentials remain in `provision.py`, so
they never enter harness arguments, logs, or MCP transcripts.
## Evidence
- `npx @ruvnet/ruview doctor --group firmware` verified all four checked-in
`release_bins` bundles (c6-adr110, c6-onboarding, s3-adr110,
s3-fair-adr110) against their `SHA256SUMS.txt` in this change's development
container.
- The write and preflight paths are covered by tests with injected fake
hardware: mismatch refusal, probe failure, flash failure, boot-log
validation, and a silent log that yields no validation.
- **No real board was flashed** during this change, so real-silicon validation
is outstanding.
## Consequences
- Operators on any OS get one verified command. Agents can review plans
without hardware authority.
- The workflow depends on Python, esptool, and pyserial. `doctor` reports each
one with a fix.
- Follow-up: an ADR-028 witness record could store the boot-log summary and
image digests.
@@ -0,0 +1,91 @@
# ADR-371: Harness training runner and mean-pose evidence gate
- **Status**: proposed
- **Date**: 2026-09-30
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: training, evaluation, evidence, pose, honesty
- **Related**: ADR-079 (camera-supervised pose), ADR-151 (room calibration),
ADR-181 (baseline-relative PCK), ADR-369
## Context
RuView has a retracted pose result (a CLAIMED 92.9%/100% that did not survive
baseline and leakage review). The project rule is that pose
PCK is quoted only with the mean-pose baseline, on a leakage-free held-out
split, with an evidence tag. The `train-pose` skill says this in prose, but
nothing enforces it. Running the trainers also needs repository knowledge:
the `tch-backend` feature, a libtorch version match, and where checkpoints go.
## Decision
### Gate (`ruview_train_gate`, pure, read-only)
The input is a small report: metric, model and baseline score, split,
train/test subjects, train end and test start, `n_test`, reproducer, and data
source.
**Hard failures:**
- no baseline;
- a random-frame split, or any split outside chronological / blocked-gap /
grouped-subject / grouped-session;
- subject overlap in a grouped-subject split;
- missing or overlapping time bounds for chronological or blocked-gap splits;
- no improvement over the baseline.
**Warnings:**
- subjects not reported;
- subjects overlapping in non-subject splits;
- `n_test` < 100;
- a near-perfect score: at or above the `implausible_score` threshold in
`src/training.js`.
**Evidence tag:** `SYNTHETIC` when `data: synthetic`, `MEASURED` only with a
reproducer, otherwise `CLAIMED`.
On PASS the gate returns the one acceptable sentence:
`Held-out <metric> X% vs Y% mean-pose baseline = +Z pp (<tag>, …, <split> split).`
### Runner (`ruview_train`, `ruview_train_plan`)
Modes:
- `pose-smoke`: `wifi-densepose-train` `train --dry-run` on the synthetic
dataset.
- `pose`: `--data-dir`, MM-Fi recordings.
- `room`: `wifi-densepose train-room`, or `cargo run -p wifi-densepose-cli`
when the binary is absent.
Rules:
- All caller paths (config, data, checkpoints, enrollment, output) must
resolve inside the trusted checkout. Realpaths are checked for inputs that
already exist.
- Pose runs use `cargo run --manifest-path v2/Cargo.toml` with the working
directory set to the checkpoint directory, default
`v2/target/ruview-train/<mode>`. The trainer's relative `logs/` therefore
never lands in the source tree.
- Without `confirm` it returns a plan. MCP requires `workspace-write` plus
`confirm`.
- Failures are classified with remedies:
- libtorch absent;
- libtorch version mismatch (tch 0.24 expects torch 2.11.0);
- link failure.
## Evidence
- **SYNTHETIC** (`npx @ruvnet/ruview train --mode pose-smoke --samples 16
--confirm`, CPU PyTorch 2.11.0, `LIBTORCH_USE_PYTORCH=1`, Linux x64): the
real trainer compiled, ran 20 epochs with early stopping, and wrote
safetensors checkpoints and `logs/training_log.csv` under
`v2/target/ruview-train/pose-smoke`. The trainer labels its own smoke
metrics "DO NOT REPORT". None are reported here.
- With PyTorch 2.14.1 the tch 0.24 C++ shim failed to compile. The runner
classified this as a version mismatch rather than a missing install.
- The gate is covered by `test/training.test.mjs`, including the leakage
patterns behind the retraction.
## Consequences
- A number can pass through the harness only with its baseline and split
attached. The gate does not verify the arithmetic that produced the input;
reviewers must still re-run the reproducer.
- Training needs the Rust toolchain and a matching libtorch. `doctor` covers
Rust; libtorch is detected at run time.
@@ -0,0 +1,66 @@
# ADR-372: Structured debugging doctor
- **Status**: proposed
- **Date**: 2026-09-30
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: diagnostics, onboarding, support, mcp
- **Related**: ADR-369, ADR-370, ADR-371, docs/TROUBLESHOOTING.md
## Context
Most failed RuView setups come down to a missing driver, pyserial, esptool, a
wrong port, uninitialized submodules, a missing wasm32 target, or a sensing
server that is not running. The old `doctor` checked only the harness itself,
gave no remedies, and had no machine-readable output for agents.
## Decision
`src/doctor.js` returns
`{ok, summary{pass,warn,fail,skip}, checks[{group,id,status,detail,remedy}], nextSteps}`.
Groups:
| Group | Checks |
|---|---|
| `runtime` | Node ≥ 20 |
| `harness` | packaged-file manifest integrity, claim-guardrail self-test, reviewed brain load |
| `hosts` | `claude` / `codex` on PATH |
| `repo` | checkout, workspace submodules |
| `rust` | cargo, wasm32 target (run inside `v2/` so the pinned toolchain is used), `wifi-densepose` binary |
| `python` | interpreter, pyserial, esptool version |
| `firmware` | checksum verification of every `release_bins` bundle |
| `serial` | pyserial port enumeration; optional named port; optional chip probe |
| `sensing` | optional sensing-server `/health` |
| `kernel` | optional `@ruvnet/ruview-kernel` WASM/napi status |
Rules:
- `fail` is reserved for a requested capability that cannot work. Absent
optional tooling is `warn`. Every non-pass check carries a remedy. The CLI
exit code is non-zero only on `fail`.
- The doctor is read-only. Two exceptions exist, both CLI/SDK options that are
absent from the MCP schema:
- `--probe` runs `esptool chip_id`, which resets the board but writes
nothing.
- `--url` makes an HTTP(S) request.
- Output is human-readable by default. `--json` gives the structured report.
`--group a,b` limits the scope.
## Evidence
Development container (Linux x64, Node 22):
- Before `pip install pyserial esptool`, the doctor reported them as `warn`
with the exact pip commands.
- After installation it showed `esptool 5.4.0` and enumerated `/dev/ttyS0`.
- All four firmware bundles verified.
- It caught manifest drift during development, and a missing wasm32 target
when run outside `v2/`. That second case led to the `v2/` working-directory
rule.
## Consequences
- Support threads can ask for `npx @ruvnet/ruview doctor --json`, and agents
can act on `nextSteps`.
- Checks run real subprocesses, so a full run takes seconds; `--group` keeps
targeted runs fast.
@@ -0,0 +1,244 @@
# ADR-373: Host device access layer — ESP32, mmWave, LiDAR
- **Status**: proposed
- **Date**: 2026-09-30
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: hardware, esp32, mmwave, lidar, serial, udp, mcp, least-authority
- **Related**: ADR-018 (CSI frame), ADR-063 (mmWave fusion), ADR-320
(sensor HAL), ADR-369 (npm operator surface), ADR-370 (flashing),
ADR-374 (remote hosts), `integrations/iphone-lidar`
## Context
Operators and agents need to answer basic hardware questions, such as "is the
node streaming?", "is the radar talking?" and "does the LiDAR see the room?",
before calibration, fusion or training. Until now each modality had its own
tool:
- the ESP32 serial monitor;
- `scripts/mmwave_fusion_bridge.py`;
- the iPhone LiDAR web relay;
- the sensing server itself.
None of these was reachable through the npm harness, its MCP server or its
SDK.
The package must stay free of runtime dependencies (ADR-263). Node has no
serial API. Hardware access is also authority: an MCP client should not get
it by default.
## Decision
Add `src/devices/` to `@ruvnet/ruview` (0.7.0) with four read-oriented tools.
| Tool | Transport | What it returns |
|---|---|---|
| `ruview_devices_scan` | pyserial port list | ports with USB VID:PID, bridge chip, likely roles (esp32 / mmwave / rplidar), and the command that confirms each role |
| `ruview_esp32_capture` | `node:dgram` UDP, receive-only | per node: packet counts by kind (magics `0xC5110001`–`0xC5110007`), CSI rate, sequence loss, RSSI, antennas/subcarriers/frequency, latest device vitals |
| `ruview_mmwave_read` | serial | MR60BHA2 (60 GHz, 115200) and LD2410 (24 GHz, 256000) frames, checksum errors, presence, distance, device-reported heart and breathing rate |
| `ruview_lidar_read` | serial or WebSocket | RPLIDAR SCAN summary (points, revolutions, range, angular coverage), or iPhone relay depth statistics |
### Rules
1. **Firmware-identical parsing.**
- The MR60BHA2 and LD2410 parsers are ports of
`firmware/esp32-csi-node/main/mmwave_sensor.c`: same SOF, `~XOR` header
and data checksums, 30-byte payload cap, and F4F3F2F1…F8F7F6F5 framing.
- ESP32 packets are parsed as the little-endian structs in `csi_collector`
and `edge_processing.h`.
- Host and node therefore agree on what a valid frame is.
2. **Serial without dependencies.** A fixed Python script (the pyserial
"pump") reads the port for a bounded time and number of bytes.
- Port, baud, duration and command bytes are passed as argv and validated
first: the port regex, a fixed baud set, and command bytes of at most 64
hex bytes.
- Output is lowercase hex, so the secret-redaction layer cannot corrupt
binary data.
- DTR control is best-effort and reported when a port does not support it.
3. **Receive-only network.**
- The ESP32 capture binds a local UDP port (bind address from a fixed set,
port ≥ 1024) and never sends to a node.
- `EADDRINUSE` is reported as "sensing server already bound".
4. **Minimal actuation.** An RPLIDAR scan sends `A5 20` (start) and `A5 25`
(stop), and drives DTR low (the A1 motor). It spins the motor but writes no
configuration. Nothing else is written to any device.
5. **No raw sensor data in results.**
- iPhone depth frames are validated with the same rules as
`integrations/iphone-lidar/web/codec.mjs` and reduced to statistics.
- The relay token comes only from `RUVIEW_LIDAR_TOKEN`. It is rejected in
the URL and never echoed.
6. **Authority.**
- All four tools require the new `device-access` MCP grant.
- The CLI and SDK run them directly, as trusted local code.
- Serial port names are pattern-checked in the schema and again in the
driver. `/dev/pts/*` is deliberately excluded: it could read another
terminal's input.
7. **Evidence.**
- A successful read is tagged `MEASURED` as a *reading* on this host.
- mmWave vitals are labelled as the radar's own estimates.
- None of these outputs is an accuracy claim.
## Evidence
Development container, Linux x64, Node 22. No physical sensors were attached.
- **ESP32 UDP**: a simulated node sent 60 ADR-018 CSI frames plus 3 vitals
packets to `127.0.0.1:5005` through `ruview esp32`.
- Result: 1 node, CSI rate 15 Hz, zero sequence loss, vitals decoded
(15.2 bpm / 69 bpm, SYNTHETIC).
- Unit test on real loopback: 2 dropped sequence numbers out of 7 produce a
loss fraction of 0.2857.
- **MR60BHA2**, emulated on a pseudo-terminal and read through real pyserial:
- auto-detected `mr60bha2`;
- 200 frames, 0 checksum errors;
- presence 1.0, distance 87 cm, 14.8 / 68.5 bpm (the emulator's values,
SYNTHETIC).
- **RPLIDAR**, emulated on a pseudo-terminal that answers `A5 20`:
- 3564 points, 99 revolutions, full angular coverage;
- the missing DTR support was reported.
- **iPhone LiDAR**, the real relay (`integrations/iphone-lidar/web/relay.mjs`)
with a simulated phone:
- 30 frames at 10 fps, median depth 1.824 m (SYNTHETIC);
- a wrong token gives `connect_failed` with a remedy.
- **Not yet done**: real ESP32-S3/C6, MR60BHA2, LD2410, RPLIDAR and iPhone
captures. Each modality needs one real capture before its parser is called
hardware-validated.
## Amendment 1 (2026-10-01): e2e suite and latency
- `harness/ruview/test/e2e/devices.e2e.mjs` (`npm run test:e2e:devices`, CI
workflow `ruview-device-e2e.yml`) automates the emulated-device runs above
through the real CLI:
- MR60BHA2 and RPLIDAR on pseudo-terminals, read via real pyserial;
- an ESP32 UDP node;
- the real iPhone relay.
- The harness still refuses `/dev/pts/*` port names. The suite links each pty
to `/dev/ttyRUVIEWE2E<n>` with root or `sudo -n`.
- `RUVIEW_E2E_REQUIRE=1` turns missing prerequisites into failures.
- mmWave auto-detect keeps its probe frames (≤ 2 s per model) and reads only
the remainder of the window. A 5 s request previously cost about 9 s.
- `ruview_lidar_read` (iPhone) gains `max_frames` and returns immediately on
a refused connection.
- MEASURED wall time for the device e2e suite on the development container:
19.6 s before these changes, 5.6 s after (concurrent cases, no redundant
re-reads).
## Amendment 2 (2026-10-01): Realtek nodes, honest capture, live analysis
Driven by a hardware run of this branch against an RTL8721Dx board (PL2303GC,
COM10) and an ESP32-C6 (CP210x, COM16) on Windows 11:
- **Realtek RAC1/RHB1 (ADR-323).** `ruview_esp32_capture` decodes the 49-byte
RAC1 envelope (8- and 16-bit tones, CRC-32 verified, SYNTHETIC flag kept)
and counts RHB1 heartbeats per sender. Before this, 3,439 live RAC1 frames
in 15 s were all reported as unknown.
- **No false success.** The capture now fails with `no_decodable_packets`
or `heartbeat_only` when packets arrive but none decode. It previously
returned `ok: true` with a MEASURED label for zero decoded packets.
`heartbeatOnlySenders` names boards that are alive but deliver no CSI — the
stall observed on a Realtek board before a reset.
- **More ESP32 packet kinds.** ADR-110 sync (`0xC511A110`) and ADR-081 mesh
envelopes (`0xC5118100`, counted by message type, never a phantom node).
A node sending several CSI shapes (the C6 sends 1x256, 1x64 and 1x128)
reports each in `csiShapes` instead of only the last one.
- **Live CSI → kernel.** `--analyze` runs the busiest (or `--node-id`) node's
dominant single-antenna shape through `@ruvnet/ruview-kernel` at the
measured arrival rate, with a bounded frame buffer (`analyze_max_frames`,
default 6,000). Results are signal-processing estimates with no reference.
- **Serial monitor no longer reboots nodes.** The monitor opens the port with
DTR/RTS deasserted. MEASURED on the C6: uptime kept rising across a monitor
run (1,612 s → 1,625 s); before, opening the port reset it (uptime ~15 s).
It takes `--baud` (Realtek logs at 1,500,000) and explains the C6/S3
USB-Serial/JTAG console when the UART is silent.
- **Device scan.** Prolific `067B:23A3` (PL2303GC) is classified `realtek`.
- **Windows.** The hosts file gets an owner-only ACL (`icacls`), failing
closed; POSIX `0o600` is ignored on Windows. The symlink-escape test skips
only where Windows refuses to create symlinks.
- **Parser speed.** Int8Array view and `sqrt` instead of per-byte reads and
`Math.hypot`: MEASURED 0.42–0.52 M → 1.33–1.67 M packets/s (about 3.2x;
three interleaved runs, median of seven each, Windows 11 x64, Node 24).
Hardware evidence (MEASURED, this host): C6 node 42, 45 s, 1,160 packets,
1,157 decoded, 0 sequence loss, 541 1x256 frames at 12.1 Hz analyzed by the
napi kernel (integrity verified). The RAC1 decoder is cross-checked against
the Rust `realtek-csi-sim` encoder (200/200 frames at 8- and 16-bit tones);
a live RAC1 re-run is pending reconnection of the Realtek board.
Real flash (MEASURED, same C6, COM16): `ruview flash --variant c6 --confirm`
with the checksum-verified `release_bins/c6-adr110` bundle wrote 4 images
(hash verified by esptool, NVS kept: node 42 rejoined). The boot log showed
13 CSI callbacks, the MGMT+DATA upgrade and no panic. After boot, `monitor`
saw CSI without a reset (uptime 36 s → 62 s) and the UDP capture decoded
183/183 packets with zero loss. The run exposed one more Windows bug, now
fixed: the boot-log capture wrote decoded text to a cp1252 stdout (the
scrubbed child env) and aborted on the firmware's U+2192 log character, so
`bootLog.captured` was false and an earlier abort would have hidden the CSI
evidence. It now writes raw bytes; a regression test forces cp1252.
Live RTL8721Dx re-run (MEASURED, board 4 on COM10, node 3, alongside the C6):
`devices` classified COM10 as `realtek`; a 20 s capture decoded 7,404/7,404
packets (6,959 RAC1 frames at 348 Hz, RSSI −38.9 dBm, 200 RHB1 heartbeats);
`monitor --baud 1500000` saw 62 CSI log lines and did not reset the board
(sequence kept rising, 80,856 → 88,257). `--analyze --node-id 3` ran 12,549
live frames at a measured 313.7 Hz through the WASM kernel (integrity
verified). The board also exposed a loss-accounting bug carried over from the
original code: one stray sequence value (+52,835, then straight back) was
counted as 52,834 lost frames, giving 97% "loss". The tracker now treats a
single out-of-window value as a stray, needs two agreeing frames to accept a
counter reset, and lets a late packet fill its own gap. Against ground truth
(unique sequence numbers over the span, 15 s): 4,783/4,783 present, true loss
0, 9 strays — the tracker reported loss 0 and 9 strays.
## Amendment 3 (2026-10-01): ESPHome radar kits and CSI buffer starvation
- **ESPHome source for radar.** The Seeed MR60BHA2 kit (a XIAO ESP32-C6 with
the 60 GHz radar) runs ESPHome, which owns the radar UART. Raw frames never
reach USB, so `mmwave --model auto` correctly returned `no_valid_frames`.
`ruview_mmwave_read` now takes `source: "esphome"` with a `host`.
- **Client:** a dependency-free, read-only client for the plaintext ESPHome
native API (TCP 6053) runs hello → connect → device info → list entities
→ subscribe states for a bounded window. It never sends a command.
- **Roles:** presence, heart, breathing, distance and target count are
found by entity name. The result uses the same summary fields as the
serial path.
- **Decoding:** ESPHome's `missing_state` flag and proto3 zero defaults are
honoured, so absent readings are excluded rather than averaged as 0.
- **Address policy:** hosts must resolve to private, link-local, CGNAT
(Tailscale) or loopback addresses.
- **Unsupported cases:** encrypted (Noise) and password-protected APIs are
reported as unsupported.
- **Live kit (MEASURED, 192.168.1.102, ESPHome 2026.9.0, project
`seeedstudio.mr60bha2_kit` `spaces-1.0`).**
- The kit had been configured for an out-of-range SSID. It was moved to
the operator's network through its own captive portal, without
reflashing.
- A 15 s read returned 154 state updates:
- presence: detected, 1 target;
- mean distance: 40.0 cm;
- device-reported heart rate: mean 75.3 bpm;
- device-reported breathing rate: mean 10.7 bpm.
- These are firmware-computed values, not validated against a reference.
- **Realtek heartbeat-only stall: root cause.** Reading the stalled board's
console without a reset showed continuous `[WLAN-W] lack of csi buf!` /
`csi buf not enough`.
- The radio produces CSI reports, but the firmware does not return report
buffers, so CSI stops while heartbeats continue.
- `ruview monitor` now counts these lines and fails with
`csi_buffer_starvation`. Observed live: 399 lines in 8 s, 0 CSI.
- The lasting fix belongs in the RTL8721Dx firmware (outside this
repository).
- **Views.** Radar results render in the terminal UI and in the console
widget (`ruview_mmwave_read` is now a UI tool). Both label vitals
"device-reported, not validated".
## Consequences
- One command per modality works across CLI, MCP and SDK, with actionable
failures (`pyserial_missing`, `port_open_failed`, `no_valid_frames`,
`port_in_use`, `connect_failed`).
- Python and pyserial become the serial dependency. `doctor` reports both.
- The package's unpacked-size ceiling rises to 320 KiB (from 224 KiB) in
both npm workflows, with the published package still free of runtime
dependencies.
- Follow-up: feed these readers into `ruview-hal` observations (ADR-320)
for fusion.
@@ -0,0 +1,78 @@
# ADR-374: Remote host access over SSH (read-only)
- **Status**: proposed
- **Date**: 2026-09-30
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: remote, ssh, hardware, least-authority, mcp
- **Related**: ADR-369, ADR-373
## Context
RuView hardware usually hangs off a different machine from the one the
operator or agent works on: a Raspberry Pi next to the ESP32s, a NUC running
the sensing server, a Cognitum Seed. Cloud agent sessions cannot reach local
USB at all.
"Host access" should not mean opening a new network service with its own
credentials: that adds attack surface and secret handling. Operators already
have SSH.
## Decision
1. **Where work runs.**
- Hardware tools always execute on the machine the hardware is attached
to.
- Interactive work runs in a local Claude Code session on that machine. It
can be driven from the Claude app through Remote Control
(`claude remote-control` in the repo).
- Other machines are reached over SSH.
2. **Hosts registry.**
- Stored in `~/.config/ruview/hosts.json` (directory 0700, file 0600).
- Written only by the CLI: `ruview hosts add --name --ssh [--port]
[--version]`. MCP can list hosts but never add them.
- Names must match `^[a-z0-9][a-z0-9-]{0,31}$`. Targets must be
`[user@]hostname`: no options, spaces, shell metacharacters or leading
dash.
3. **Transport.**
- Local ssh arguments: `ssh -o BatchMode=yes -o ConnectTimeout=10 -o
StrictHostKeyChecking=yes -T [-p N] -- <target> <command>`.
- No password prompts, and no trust-on-first-use from the tool. The
operator pins the host key once interactively.
- The remote command is `npx -y @ruvnet/ruview@<exact version> call <tool>
--read-only --args-json <json>`, with every token single-quoted for the
remote POSIX shell.
4. **Read-only, twice.**
- Locally, `ruview_host_run` accepts only tools that are read-only, use no
credentials, and are not `ruview_host_*` (no host hopping). It validates
arguments against the tool schema before sending.
- Remotely, `call --read-only` repeats the check and schema validation.
- Flash, train, calibrate and Cognitum Spaces reads are never forwarded.
5. **Grants.** Over MCP, `ruview_host_run` needs `remote-host` plus the
remote tool's own grant (for example `device-access`).
6. **Failures** are classified with remedies: `host_key_unverified`,
`ssh_auth_failed`, `host_unreachable`, `remote_node_missing`,
`unknown_host`, `remote_tool_not_allowed`.
## Evidence
Tests in `harness/ruview/test/remote.test.mjs` cover:
- host-entry injection cases (`-oProxyCommand=…`, `;`, spaces, `$(…)`,
backticks);
- quoting, verified by running the quoted command through a real `sh`;
- the refusal list;
- grant composition;
- the file mode;
- `call --read-only` refusing `ruview_node_flash` through the real CLI.
No SSH server was available in the development container, so a live
round-trip to a real host is still to be done.
## Consequences
- Remote hosts get the same tools with no new daemon, port or credential
store.
- The remote needs Node 20+ and npm registry access (or a pre-installed
version), and `ssh` must be on the local PATH. `doctor --group remote`
checks the local side.
- Remote mutations stay manual by design.
@@ -0,0 +1,202 @@
# ADR-375: MCP Apps console, HTTP transport and terminal UI for `@ruvnet/ruview`
- **Status**: proposed
- **Date**: 2026-10-01
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: npm, mcp, mcp-apps, chatgpt, ui, cli, http, least-authority, performance
- **Related**: ADR-263 (harness hardening), ADR-265 (distribution and size
budget), ADR-369 (operator surface and SDK), ADR-373 (device access),
ADR-374 (remote hosts)
## Context
The harness exposes one tool registry through the CLI, an MCP stdio server and
an SDK (ADR-369). Hardware testing of the device tools (ADR-373 Amendment 2)
showed three gaps in how results reach people and models:
- **MCP results were text only.** Every tool returned pretty-printed JSON in a
text block, with no `structuredContent`. The server declared protocol
`2024-11-05`, and `resources/list` was always empty, so MCP Apps hosts and
ChatGPT had nothing to render.
- **ChatGPT could not connect.** ChatGPT connectors reach MCP servers over
HTTPS. The harness spoke stdio only.
- **The CLI printed raw JSON to humans.** A live capture of two nodes was a
1.2 kB JSON object in the terminal, with alerts such as heartbeat-only
senders easy to miss.
Reference implementations for the target hosts (the `web-based-chatgpt-mcp-starter`
and `signal-to-swarm` sites) declare a `ui://` HTML resource and point UI tools
at it through both the MCP Apps key (`_meta.ui.resourceUri`) and the ChatGPT
keys (`openai/outputTemplate`, `openai/widgetAccessible`).
The package must stay free of runtime dependencies (ADR-263) and inside a
reviewed size budget (ADR-265).
## Decision
### 1. One console widget, two host families
`src/ui/console-widget.js` holds one self-contained HTML resource,
`ui://ruview/console-v1.html`, with mimeType `text/html;profile=mcp-app`.
- **Tools.** `ruview_esp32_capture`, `ruview_devices_scan` and `ruview_doctor`
carry `_meta.ui.resourceUri`, `ui/resourceUri`, `openai/outputTemplate` and
`openai/widgetAccessible`.
- **Data intake.** The widget reads results from either host family:
- MCP Apps: a `ui/initialize` handshake over `postMessage`, then
`ui/notifications/tool-input` and `ui/notifications/tool-result`;
- ChatGPT: `window.openai.toolOutput` and the `openai:set_globals` event.
- **Refresh.** A Refresh button re-runs the same tool with the same
arguments, through `window.openai.callTool` or an MCP Apps `tools/call`
request.
- **Safety.**
- The resource declares an empty CSP allowlist and makes no network
requests.
- Every value reaches the DOM through `textContent`; there is no
`innerHTML`.
- Messages are accepted only from the parent frame.
- A test pins all of this.
- **Design.** The visual language follows the reference sites: dark ground,
`#d0ff72` accent, monospace labels, and plain wording about what each value
is. Device vitals are labelled "device-reported (not validated)". Kernel
outputs carry the "no reference measurement" note, and SYNTHETIC frames
raise a warning.
### 2. Structured results and protocol negotiation
- **Protocol negotiation.** `initialize` negotiates `2025-06-18`,
`2025-03-26` or `2024-11-05`, and advertises `resources`.
- **Result shape.** `tools/call` returns `structuredContent` (always an
object), a text block with the same JSON in compact form, and
`_meta['ruview/tool']` set to the canonical tool name (so an aliased call
still refreshes correctly).
- **Compact text.** Compact rather than pretty-printed JSON cuts each result
by 17–33% (MEASURED on this host: guidance 17.5%, device scan 20.7%, live
capture 33.4%). Clients that read only `content` still receive the
complete result.
- **Titles.** Tools gain `title` and `annotations.title`.
`handleRpc` is transport-neutral, and `createDispatcher` holds the FIFO tool
chain both transports share (ADR-263 O2).
### 3. Streamable HTTP transport
`ruview mcp start --http [--host 127.0.0.1] [--port 8790] [--allow-origin URL]`
starts `src/mcp-http.js`.
- **Authentication on every request.** A request must carry either
`Authorization: Bearer <token>`, or the secret-path form `/mcp/<token>`.
The second exists because ChatGPT connectors cannot send a static header.
- Tokens are compared in constant time and must be at least 16 characters.
- The token comes from `RUVIEW_MCP_TOKEN`, never argv. If unset, one is
generated and printed once to stderr.
- **Least authority.**
- It binds to loopback by default.
- Browser `Origin`s are refused unless allowlisted (DNS-rebinding defence).
Server-to-server clients send no Origin.
- `workspace-write` and `hardware-write` are never honoured over HTTP, even
when granted in `RUVIEW_MCP_GRANTS`. Flashing and calibration stay
stdio/CLI-only.
- **Protocol surface.**
- POST only; single JSON responses with no SSE stream and no sessions.
- Batches and unsupported `MCP-Protocol-Version` values are refused.
- Bodies are capped at 256 KiB.
- Notifications return 202.
- `/healthz` reports only the name and version.
Exposing the port to ChatGPT (a tunnel or reverse proxy with TLS) is the
operator's choice. The startup banner says so and warns on a non-loopback
bind.
### 4. Terminal UI
`src/cli-ui.js` renders results for a TTY:
- aligned tables, with unicode rate bars;
- colour that respects `NO_COLOR` and `TERM=dumb`;
- alerts and fixes ahead of the data;
- a stderr countdown for timed captures.
Pipes and `--json` keep receiving exact JSON. `ruview call` always prints
JSON, because remote hosts parse it.
`ruview esp32 --watch [--seconds 3]` redraws repeated capture windows with a
per-node rate sparkline.
`ruview monitor` now flags Ameba CSI report-buffer starvation (`lack of csi
buf` / `csi buf not enough`). This is the root cause of the Realtek
heartbeat-only stall, observed live: 399 starvation lines in 8 s, 0 CSI.
### 5. Startup
- **Launcher split.** `bin/cli.js` is a thin launcher, and `src/cli.js` loads
the tool registry, doctor, brain, host adapters and transports on demand.
- **Measured effect.**
- `--help`/`--version`/`skills` skip the registry: about 20% faster, median
of 9 interleaved runs against the previous build on this host (MEASURED).
- The MCP path needs the whole registry to list tools, so its start time is
unchanged within noise.
- **Compile cache rejected.** Node's on-disk compile cache
(`module.enableCompileCache`) was evaluated: module load was 27.7 ms with
it and 28.5 ms without (25 interleaved runs, MEASURED). That is within
noise, so it was not adopted.
### 6. Size budget
The additions cost about 38 kB:
- widget, 12.6 kB;
- HTTP transport;
- terminal UI;
- CLI split.
The `@ruvnet/ruview` unpacked budget rises from 320 KiB to 384 KiB in
`npm-packages.yml` and `ruview-npm-release.yml`. The package is about 338 kB.
## Evidence
All on Windows 11 x64, Node 24, with an ESP32-C6 (CP210x, COM16) and an
RTL8721Dx (PL2303GC, COM10) attached. All results MEASURED unless noted.
**Browser e2e.** A real MCP Apps host flow in headless Chrome:
- The host lists tools and reads the `ui://` resource over HTTP MCP, then
loads it in a sandboxed iframe (`allow-scripts` only).
- It answers `ui/initialize` and delivers a live `ruview_esp32_capture`
result.
- The widget's Refresh button triggered a second real capture through the
host (108 → 105 packets).
- The devices and doctor views rendered, in dark and light themes. The only
console error was the test host's own missing favicon.
**Security probes** against the live HTTP server:
- no token: 401;
- wrong token: 401;
- foreign Origin: 403;
- GET: 405;
- `ruview_node_flash` with `hardware-write` granted in the environment:
`authority_denied`.
**Tests.**
- `npm test`: 160 pass, 1 skip.
- `test:security`: 83 pass, 1 skip.
- New suites: `mcp-apps.test.mjs` (protocol, metadata, widget safety, HTTP
auth/origin/size/write denial) and `cli-ui.test.mjs` (renderers, colour
rules, the starvation detector run against a stand-in serial module).
**Tarball smoke** from a clean install:
- CLI and SDK work;
- stdio MCP negotiates `2025-06-18` and serves the widget and structured
results;
- HTTP MCP lists 22 tools, 3 with `ui://` metadata.
## Consequences
- MCP Apps hosts and ChatGPT render node streams, device scans and
diagnostics instead of raw JSON, and can refresh them in place.
- Remote MCP clients can use the read tools over authenticated HTTP. Write
tools remain local by construction.
- Each tool result costs fewer model tokens.
- The widget is a second place where result shapes are interpreted. Changes
to the capture, devices or doctor result shapes must update it, and its
version suffix (`console-v1`) changes when the contract does.
- Host behaviour for MCP Apps and ChatGPT keys is followed as documented by
the reference implementations. A live ChatGPT connector session (public
HTTPS URL) has not been run yet: CLAIMED, pending.
@@ -0,0 +1,142 @@
# ADR-376: `ruview` — one npm install for every RuView component
- **Status**: proposed
- **Date**: 2026-10-01
- **Deciders**: rUv (requested and approved the release); RuView maintainers
- **Tags**: npm, packaging, umbrella, mcp, homecore, kernel, release
- **Related**: ADR-263 (harness hardening), ADR-265 (distribution, D2 CI-only
publishing, D4 bin ownership), ADR-285 (Homecore metaharness), ADR-368
(kernel), ADR-369 (operator surface), ADR-373 (device access), ADR-375 (MCP
Apps console, HTTP transport, terminal UI)
## Context
RuView is positioned as an ambient intelligence platform rather than a sensor.
Its sensing is ahead of comparable packaged options; packaging is the
bottleneck. Today the capabilities are split across three packages:
| Package | Contents | Published before this ADR |
|---|---|---|
| `@ruvnet/ruview` | operator harness | 0.5.1 |
| `homecore` | Homecore developer metaharness | never |
| `@ruvnet/ruview-kernel` | WASM vitals kernel | never |
The unscoped name `ruview` was unclaimed. A user has to know all three
packages, install them separately, and run three CLIs and two MCP servers.
## Decision
1. **The umbrella.** Publish an unscoped `ruview` package
(`harness/ruview-umbrella`). It depends on exact versions of the three
component packages and adds no capability of its own:
- **CLI.** One binary, `ruview`:
- `ruview homecore …` → the Homecore CLI;
- `ruview kernel <doctor|info|selftest|parity|synth|analyze|bench>` → the
kernel CLI;
- bare `ruview kernel [--backend …]` keeps its harness meaning (the
SYNTHETIC self-test);
- `ruview capabilities` lists every component, version and tool;
- everything else goes to the `@ruvnet/ruview` CLI unchanged.
- **MCP server.** One server, `ruview mcp start [--http]`. It serves the
harness tools (`ruview_*`, including the `ui://` console) plus the
Homecore tools (`homecore_*`, already prefixed, so no collision).
- Non-Homecore traffic goes to the harness handler, so the harness
authority model, grants and HTTP rules (ADR-375) apply unchanged.
- Homecore calls go to Homecore's `runTool`, with Homecore's own
policy.
2. **Bin ownership (amends ADR-265 D4).** The `ruview` command name moves to
the umbrella. `@ruvnet/ruview`'s bin is renamed `ruview-harness`.
- **Why:** a clean-directory tarball install showed the collision. With
both packages declaring `ruview`, `node_modules/.bin/ruview` linked to
the harness CLI, so `ruview capabilities` and `ruview homecore` failed.
Which package wins that link is not guaranteed.
- **Unaffected:** `npx @ruvnet/ruview …` still works, because npx runs a
package's only bin whatever its name. The umbrella delegates every
non-umbrella verb to the harness, so `ruview <verb>` behaves as before.
- **Changed:** a global install of `@ruvnet/ruview` alone now provides
`ruview-harness`.
3. **Harness changes needed to embed it** (no new runtime dependencies):
- **New exports:** `./cli`, `./cli-ui`, `./mcp`, `./mcp-http`, `./ui`,
`./esphome` and `./kernel`.
- **Injectable MCP handler:** `run(args, { mcpHandler })` and
`startMcpServer({ handler })`, so the merged server reuses the stdio and
HTTP transports.
- **Kernel declared as an optional peer:** `@ruvnet/ruview-kernel`.
- **`setKernelImporter`:** lets the umbrella resolve the kernel from its
own dependency tree. Node resolves bare specifiers from the importing
file's real path, which misses a sibling dependency under linked or
pnpm-style installs.
- Found while testing: `kernel_not_installed` through a linked umbrella.
- Only embedding code can call it; MCP arguments cannot choose the
module.
- **Homecore** gains a `./cli` export; **the kernel** gains `./cli`.
4. **Versions for this release.**
| Package | Version | Notes |
|---|---|---|
| `@ruvnet/ruview` | 0.8.0 | 0.7.0 was never published |
| `homecore` | 0.1.0 | first publish |
| `@ruvnet/ruview-kernel` | 0.1.0 | first publish; WASM only |
| `ruview` | 0.8.0 | umbrella |
The kernel ships WASM only. The napi backend stays optional and absent,
because a workstation build would cover only win32-x64. `loadKernel({
backend: 'auto' })` reports the honest fallback.
5. **Publishing exception (one-off, amends ADR-265 D2 for this release).**
- rUv explicitly chose a local `npm publish` from the operator
workstation (ruvzen) over the CI-only release workflow.
- Packages are published in dependency order: kernel → homecore →
@ruvnet/ruview → ruview.
- Each is published only after its full gate passes locally (tests,
security tests, manifest, version-literal and pack-content/size gates),
plus a clean-directory tarball install and smoke test of the umbrella.
- These versions carry no npm provenance attestation.
- Subsequent releases return to the CI workflow. Adding `ruview` and
`@ruvnet/ruview-kernel` to `ruview-npm-release.yml` is follow-up work.
## Release outcome (2026-10-01)
**Published from ruvzen** (exact tarballs that passed the clean-directory
smoke test; sha256 in the release thread):
| Package | Version |
|---|---|
| `@ruvnet/ruview-kernel` | 0.1.0 |
| `homecore` | 0.1.0 |
| `@ruvnet/ruview` | 0.8.0 |
All three were re-installed from the registry and smoke-tested:
- `npx -y @ruvnet/ruview@0.8.0` works;
- the harness self-test resolves the published kernel;
- a live ESPHome radar read succeeded.
**`ruview` was refused by npm:** `403 Forbidden — Package name too similar to
existing package iview` (typosquat protection).
- The umbrella is ready in `harness/ruview-umbrella`. Until npm support grants
the name, it is unpublished.
- If the request is declined, the fallback is a scoped name
(`@ruvnet/ruview-platform`).
## Consequences
- `npx ruview` is the single entry point for operators and agents. One MCP
connector gives an agent, or ChatGPT, every RuView and Homecore tool.
- The component packages stay independently installable and testable.
`@ruvnet/ruview` remains free of runtime dependencies.
- The umbrella pins exact component versions, so each component release needs
an umbrella release to reach `npx ruview` users.
- The local publish trades provenance for speed for this release only. The
exception is recorded here and in the #swarm coordination thread.
## Evidence
- **Umbrella tests:** 4/4, covering CLI dispatch to all three components,
merged MCP tool listing and routing, the harness authority model through
the merged server, and stdio and HTTP transports.
- **Component gates:**
- harness: 166 tests, 89 security tests;
- Homecore: 27 tests, 23 security tests;
- kernel: 21 tests, 3 skipped native cases; 12 security tests;
- all manifests verified, 0 audit findings.
- **Clean-directory tarball smoke:** recorded in the release commit and the
#development post.
@@ -0,0 +1,101 @@
# ADR-377: `ruview-live` — a Claude Code mod shipped inside `@ruvnet/ruview`
- **Status**: proposed
- **Date**: 2026-10-01
- **Deciders**: rUv (requested integration into the npm package); RuView maintainers
- **Tags**: claude-code, mods, function-hooks, ui, npm, least-authority
- **Related**: ADR-373 (device access), ADR-375 (MCP Apps console, terminal UI),
ADR-376 (umbrella package); Claude Code mods (function hooks): the
claude-code issue 91870 and the `mods/` built-ins (`diff` is the closest
analogue)
## Context
Claude Code mods are plugins whose behaviour is a function-hooks module: one
`register(on, options)` that hooks engine events as `($, e, next)`. They can:
- register slash commands;
- open panes beside the transcript and draw them with the surface's elements;
- set a status line;
- run processes through `$.process.run`.
That fits RuView's goal of putting the live state of the sensors where an
operator or agent already works.
RuView already has:
- a read-only CLI with `--json` results (ADR-373, ADR-375);
- an MCP server and the ChatGPT/MCP Apps console;
- a plugin marketplace in this repository.
## Decision
1. **Ship a mod, `ruview-live`, inside `@ruvnet/ruview`** (`mod/`, packaged
files `mod/.claude-plugin/`, `mod/hooks/`).
- It is listed in the repository marketplace as `ruview-live@ruview`, with
source `./harness/ruview/mod`.
- One copy serves both installs: the npm package and the marketplace.
2. **Behaviour.**
- **`/ruview`** toggles a pane (`ruview-live`). While it is open, a
`$.clock.every` timer (default 15 s, minimum 5 s) runs one read-only
capture.
- **`/ruview refresh`** runs one capture and returns the summary.
- **`/ruview off`** closes the pane.
- **The pane** shows nodes (rate, loss, RSSI, shape, SYNTHETIC flag), the
optional ESPHome radar (presence, distance, device-reported vitals), and
alerts, including heartbeat-only boards with the CSI-buffer-starvation
hint. It has Refresh (`r`) and Close (`c`) buttons.
- **The status line** is set on every refresh.
3. **Data comes only from the harness CLI beside the mod.**
- The mod runs `node <plugin root>/../bin/cli.js esp32 … --json` and
`mmwave --source esphome … --json`, as fixed argv arrays, through
`$.process.run`. So the pane shows what the tested tools return.
- Mods may import nothing but their own files and `claude-code` (the
validator refuses `node:url`). The CLI is therefore located through
`$.plugin.root`.
4. **Least authority.**
- The mod never runs a write verb (flash, calibrate, train). A unit test
pins the argv.
- The radar host is validated before use.
- Options are bounded: ports 1024–65535, refresh ≥ 5 s, capture 1–10 s.
5. **Discovery.** `ruview mod` (and `ruview mod --json`) prints the installed
path, the `claude --plugin-dir` command, the marketplace commands, usage,
settings, and the early-access note.
## Evidence
All on Claude Code 2.1.286, Windows 11 x64.
- **Validation:** `claude plugin validate harness/ruview/mod` passes. It lists
hooks `session.start`, `command.run{ruview}`, `ui.render{Pane}`,
`ui.close{ruview-live}` and `session.end`, and calls `$.clock.every`,
`$.clock.now`, `$.command.register`, `$.process.run` and `$.ui.*`.
`claude plugin validate .` passes for the marketplace.
- **Engine tests:** `claude plugin test harness/ruview/mod`, with
`CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1`, passed 2/2 in three consecutive
runs. They check:
- `/ruview` opens the pane and runs the exact read-only argv;
- the status reads `RuView · 2 nodes · 1 alert`;
- the timer refreshes while open and **never after close**;
- an honest `no_packets` status.
- **Harness unit tests:** `test/mod.test.mjs`, 6/6, under plain Node:
settings bounds, host validation, argv, failure parsing, model/status, pane
drawing and button wiring, and the no-imports rule.
- **Real engine, headless (MEASURED):**
`claude --plugin-dir ./mod -p "/ruview refresh"` answered
`RuView · 0 nodes · 1 alert` (no CSI node was streaming). With `radarHost`
set to the live Seeed MR60BHA2 kit, it answered
`RuView · 0 nodes · radar clear · 1 alert`.
- **Note:** under Git Bash a leading `/ruview` is rewritten into a Windows path
unless `MSYS_NO_PATHCONV=1` is set. This is not an issue inside the Claude
Code prompt.
## Consequences
- An operator sees node and radar health beside the conversation without
asking for it. Agents see the same facts through MCP.
- **Mods are early access.** The API may change between Claude Code releases,
and hooks modules load only where function hooks are enabled. The mod is
therefore optional, and nothing else in the package depends on it.
- **Trust.** Like every mod, it runs inside Claude Code with Claude Code's
access. It is read-only by construction, and the README says to install it
only from a trusted source.
- A refresh binds UDP 5005 for the capture window. While the sensing server
holds the port, the pane reports `port_in_use` instead of reading.
@@ -0,0 +1,144 @@
# ADR-378: `ruview-live` showcase views — CSI waterfall, radar fan, animated cells
- **Status**: proposed
- **Date**: 2026-10-01
- **Deciders**: rUv ("make this a showcase … the art of the possible for mods"); RuView maintainers
- **Tags**: claude-code, mods, ui, raster, animation, csi, radar, honesty
- **Related**: ADR-377 (the `ruview-live` mod), ADR-373 (device access),
ADR-375 (terminal UI), ADR-018 / ADR-323 (CSI formats)
## Context
ADR-377 shipped `/ruview`: a pane of cards built from text and sparklines. The
mod engine can do much more on the terminal surface:
- `Raster` is a grid of cells, each with a glyph and a 24-bit foreground and
background.
- `$.ui.blit` repaints a mounted Raster in place, up to 60 frames a second,
without redrawing the tree.
- Buttons take hotkeys.
The pane only showed summaries. It never showed the signal itself: per-subcarrier
CSI amplitude over time, or the radar's range as geometry.
## Decision
### Three views, keys `1` `2` `3`
| View | What it draws | Data |
|---|---|---|
| Overview | the ADR-377 cards | unchanged |
| CSI waterfall | subcarrier bins across, time down; two frames per text row through `▀` (foreground = upper frame, background = lower); 5th–95th percentile colour limits with a colour bar | `ruview esp32 --spectrum` (new) |
| Radar | top-down 120° fan: range rings, a fading trail of past distances, the current distance as an arc; braille line charts of heart, breathing and distance | ESPHome radar kit (ADR-373) |
`n` cycles nodes in the waterfall when more than one streams. The live views
refresh every `liveRefreshSeconds` (default 4, range 2–60); the overview keeps
`refreshSeconds`.
### Harness: `--spectrum`
`captureEsp32` and the `ruview_esp32_capture` tool take three new arguments:
- `spectrum`: boolean;
- `spectrum_bins`: 8–128, default 48;
- `spectrum_frames`: 8–256, default 64.
A `SpectrumCollector` keeps a ring of the newest frames for each node and CSI
shape. It averages each frame's amplitudes into equal bins, and reports the
dominant shape per node with its arrival rate and `synthetic` flag. The
collector is read-only and bounded (8 nodes), and it adds nothing to a capture
that does not ask for it.
### Animation with `$.ui.blit`
One 80 ms timer (about 12 fps) repaints only these keyed Rasters:
- **shimmer**: a highlight sweeping along the rule under the title. Pure
decoration.
- **pulse**: a ♥ that beats at the device-reported heart rate and a gauge that
fills and empties at the device-reported breathing rate. It is labelled *a
metronome, not a waveform*.
- **fan**: a sonar ring that leaves the sensor and reaches the measured range.
- **waterfall replay**: frames arrive a capture at a time, so the view drains
them at their measured arrival rate instead of jumping every refresh. It is
labelled *replayed at the arrival rate*.
The first paint and every frame come from the same function (`picturesOf`). A
test asserts that each picture keeps its size across frames, because a blit of
another size is refused.
Blits are fire-and-forget: a blit resolves only once a frame is painted, and
the surface folds blits between frames anyway.
### Honesty rules (CLAUDE.md)
- The waterfall says **MEASURED — live CSI amplitude received on this host** or
**SYNTHETIC — simulator frames**, from the packets' own flag.
- The radar view states *range only: this kit reports distance, not bearing*.
The arc covers every bearing at that range; nothing implies a position.
- Vitals stay *device-reported, not validated*. Values outside 40–180 bpm
(heart) or 4–40 bpm (breathing) are flagged.
- Chart scales fit the data so a rhythm is visible. The plausible band is
shaded only where it overlaps the scale.
- Surfaces without `Raster` (desktop, VS Code, mobile) get a note in the
showcase views instead of a tree the surface would refuse. The overview works
everywhere.
- Only nodes in the latest capture stay in the waterfall. A node that stops
streaming, or a failed capture, clears its frames, so old frames are never
shown as live. Frames join across captures only when the CSI shape and bin
count both match.
- Multi-antenna ESP32 nodes are drawn from their first chain. ADR-018 lays the
tones out antenna-major; this applies to the waterfall only, and the vitals
kernel still takes single-chain frames.
- Animation runs on the real clock, the same one the full render reads. The
pulses keep the reported rates even when a frame fires late.
### Structure
Mods may import only their own files, so the hooks module is split into:
- `model.mjs`: settings, argv, results, model and history (pure);
- `raster.mjs`: Grid, base64, colour map, waterfall, braille, radar fan, line
chart (pure);
- `anim.mjs`: frame functions of time (pure);
- `views.mjs`: drawing;
- `register.mjs`: wiring, timers and blits. It re-exports the rest so tests
import one module.
## Evidence
- **MEASURED** (real silicon, 2026-10-01):
- Hardware: an ESP32-C6 on COM11, flashed with the hardware-verified
`release_bins/c6-adr110` image (v0.7.0; SHA-256 checked) and provisioned to
ruv.net, node 6.
- Boot log: `Got IP: 192.168.1.103` and `UDP sender initialized:
192.168.1.45:5005`.
- Captures: `ruview esp32 --spectrum` received 29–50 frames per 2 s at 7–18
Hz and −37 to −78 dBm, in 1×64 and 1×256 shapes.
- Waterfall: a 1×256 capture rendered as 114 bins showing the occupied band
against guard and null tones.
- Reproducer: `npx @ruvnet/ruview esp32 --seconds 4 --spectrum --json` with
a node targeting the host.
- **MEASURED** (device-reported): the Seeed MR60BHA2 kit at 192.168.1.102 drives
the radar view.
- **SYNTHETIC**: the `realtek-csi-sim` simulator exercises the waterfall in
development; its frames carry the synthetic flag and the view says so.
- **Tests**:
- 11 mod tests and 11 showcase tests: Grid/base64 against Node's encoder,
colour map monotone, waterfall ordering, braille bits, fan scaling and
ping, chart scaling, frame functions, constant picture sizes, MEASURED and
SYNTHETIC labels, surfaces without Raster;
- a loopback test of `--spectrum`;
- 4 engine tests (`claude plugin test mod`), including the waterfall tab
sending `--spectrum`, drawing the Raster and animating it with blits.
## Consequences
- The package grows by about 37 KB. The reviewed unpacked ceiling rises from
384 KiB to 448 KiB in `npm-packages.yml` and `ruview-npm-release.yml`. There
are still no runtime dependencies.
- The live views poll more often: a 2 s capture every 4 s by default. A
capture binds the UDP port briefly, so a sensing server on the same port
still reports `port_in_use`, as in ADR-377.
- The animation timer runs only while the pane is open, and stops on close and
session end.
@@ -0,0 +1,199 @@
# ADR-379: An OTA'd image confirms itself with a bounded health check, or rolls back
**Status:** Proposed. Hardware-verified on 2026-09-29 on an ESP32-S3 (node 4) and an ESP32-C6 (node 5).
**Date:** 2026-09-29
**Numbering:** On 2026-10-02 `main` had up to ADR-378 and open PRs claimed
nothing higher. If another branch also takes 379, renumber whichever merges
second.
## Context
Every 16 MB node was flashed with `CONFIG_BOOTLOADER_APP_ROLLBACK_ENABLE=y`
(`sdkconfig.defaults.16mb`). With that option the bootloader starts a freshly
OTA'd image in `ESP_OTA_IMG_PENDING_VERIFY`. The app must then call
`esp_ota_mark_app_valid_cancel_rollback()`. If it doesn't, the bootloader marks
the image `ABORTED` at the next reset and boots the previous slot.
Nothing in `firmware/esp32-csi-node/main` made that call, and `nm` on the ELF
confirmed the symbol was not linked. RUNBOOK.md said an
`ota_rollback_boot_check()` "sits in the app", but no such function existed
anywhere in the tree. The consequences:
- An image delivered by `POST /ota` ran until the next reset, then reverted
silently. OTA updates were not durable.
- While the image was pending, `esp_ota_begin()` refused a second OTA with
`ESP_ERR_OTA_ROLLBACK_INVALID_STATE`.
- Nothing checked health, so a bad image was never rolled back on purpose. It
only reverted if it happened to crash or be power-cycled.
A second defect meant none of this was ever reached: **OTA never succeeded
on this firmware before this change.** MEASURED on 2026-09-29, on node 4 (an
ESP32-S3 running the v0.8.12 16 MB build, with the OTA key provisioned):
- A `POST /ota` of a valid 1,228,672 B image logged `OTA update started`, then
`stack overflow in task httpd`, then `rst:0xc RTC_SW_CPU_RST`. The node came
back on its old slot. Both attempts failed the same way.
- The cause: the server used `HTTPD_DEFAULT_CONFIG()`, which gives a 4096 B
task stack. The handler held a 1 KB receive buffer on that stack, while
calling `esp_ota_*` and formatting a float progress log.
## Decision
At boot, `ota_health_start()` (in `main/ota_health.c`, called early in
`app_main`) reads the running partition's state:
| state | meaning here | action |
|---|---|---|
| `PENDING_VERIFY` | just OTA'd, rollback armed | run the health check |
| `VALID` | USB-flashed with `ota_data_initial.bin` (the bootloader writes VALID), or already confirmed | nothing |
| `UNDEFINED` | selected by a build without rollback; boots unconditionally | nothing |
| `NEW` | bootloader lacks rollback, so it never promoted the image | nothing |
| error (`NOT_FOUND`/`NOT_SUPPORTED`) | no otadata entry, or factory partition | nothing |
For this table I read the ESP-IDF v5.5 source: `bootloader_utility.c`
(`set_actual_ota_seq`, the NEW to PENDING_VERIFY to ABORTED transitions) and
`esp_ota_ops.c`.
**The health check.** A low-priority task polls once a second and feeds
`ota_health_step()`, which is a pure function in `ota_health.h`:
- **Signals, both latched:** the STA got an IP (`IP_EVENT_STA_GOT_IP`, via
`ota_health_note_got_ip()` in `main.c`'s handler), and the stream sender
accepted at least one CSI frame (`csi_collector_get_send_ok_count() > 0`).
The second signal covers driver callback, serialization and `sendto()`
together.
- **Minimum uptime** `CONFIG_OTA_HEALTH_MIN_UPTIME_S` (default 30 s): the image
is not confirmed before this point even with both signals present. An image
that crashes soon after starting its pipelines still reverts, because a crash
while pending is a rollback.
- **Deadline** `CONFIG_OTA_HEALTH_TIMEOUT_S` (default 120 s, measured from
boot). Both signals are required by then. The minimum uptime is clamped to
the deadline.
- **Pass:** `esp_ota_mark_app_valid_cancel_rollback()`, which logs
`marked valid after <reason> in <ms> ms`.
- **Fail:** logs `health check failed: <reason>, rolling back`, then calls
`esp_ota_mark_app_invalid_rollback_and_reboot()`. If no other bootable slot
exists, that call returns an error. The node then logs and stays up rather
than reboot-looping.
- **Terminal verdict:** once the check has confirmed or rolled back, it never
changes.
**Negative-test hook.** `CONFIG_OTA_HEALTH_FORCE_FAIL` (default n, test only)
makes the check roll back at the point where it would have confirmed. Use it to
prove rollback end to end.
**Remote visibility.** `GET /ota/status` now includes `ota_state` (`valid`,
`pending_verify`, `new`, `undefined`, `other`, or `none`). An operator can
confirm durability without a serial console.
**The httpd stack.** The OTA/WASM HTTP server now sets `config.stack_size =
CONFIG_OTA_HTTPD_STACK_SIZE`, which defaults to 8192. Three smaller changes go
with it:
- The OTA receive buffer is static rather than on the stack. Handlers run one
at a time on the single httpd task, so one buffer is enough.
- The progress log prints an integer percent instead of a float.
- `GET /wasm/list` keeps its 2 KB JSON buffer on the heap, not the stack.
`recv_wait_timeout` stays at 30 s. Each `POST /ota` and `POST /wasm/upload` now
logs `httpd stack after <handler>: N of M bytes never used`, so the size can
be re-tuned from device measurements.
Sizing: this is an estimate from static data, not a device measurement. It
comes from a `-fstack-usage` build of the S3 16 MB configuration with IDF v5.5.
- **Frame sizes along the OTA path** (bytes):
| function | bytes |
|---|---|
| `httpd_thread` | 144 |
| `httpd_sess_process` | 32 |
| `httpd_req_new` | 176 |
| `httpd_uri` | 64 |
| `ota_upload_handler` | 192 (1,216 with the old 1 KB buffer) |
| `esp_ota_end` | 48 |
| `image_validate` | 304 |
| `esp_image_verify` | 32 |
| `image_load` | 80 |
| `process_segments` | 112 |
| SHA-256 | about 110 |
| mmap/flash read | about 190 |
- That chain comes to about 1.5 KB now, and about 2.5 KB with the old buffer.
- Two things are not in the static data:
- libc's `vfprintf` behind every `ESP_LOG`. This build uses full newlib,
not nano, and the old progress line formatted a float.
- The flash-chip driver, which is reached through function pointers.
Together they plausibly add 1 to 1.5 KB. That explains an overflow at 4096
with the old handler.
- The WASM upload path adds `wasm_upload_process` (336) and
`rvf_verify_signature` (352), plus a wasm3 module load that wasn't measured.
- 8192 leaves roughly 4 to 5 KB above the estimated worst case, and costs
4 KB of internal RAM. The high-water-mark log is the check on that estimate.
When the build has no `CONFIG_BOOTLOADER_APP_ROLLBACK_ENABLE` (the 8 MB S3 and
4 MB C6 CI lanes), both entry points compile to no-ops.
## Consequences
- OTA updates become durable on rollback-enabled nodes once the image passes
the check. A second OTA is accepted after confirmation, which can come as
early as `MIN_UPTIME_S`.
- A bad image now reverts instead of limping, but "bad" only means it failed
to reach the network or deliver CSI. An image that passes and then
misbehaves later is not caught.
- If the AP is down when a node boots an OTA'd image, a good image rolls back.
That is the safe direction: push it again later.
- Mock/QEMU builds that skip Wi-Fi can never pass. They must not enable
bootloader rollback, and none do today.
- Bootloader capability is still invisible remotely (RUNBOOK §2). A node
whose bootloader lacks rollback reports `ota_state: "new"` after an OTA. That
is the first remote hint of which bootloader a board has.
## Evidence
- Host test `firmware/esp32-csi-node/test/test_ota_health.c` runs in
`make host_tests`, which CI runs. It covers pass, each timeout reason, the
terminal verdict, force-fail, the clamp, and zero soak. SYNTHETIC: it
exercises the decision function, not a device.
- ESP-IDF v5.5 S3 and C6 builds with rollback enabled link
`esp_ota_mark_app_valid_cancel_rollback` (verified with `nm`). A build is not
hardware evidence.
- The httpd overflow is MEASURED on hardware (node 4, above).
- Hardware verification, MEASURED on 2026-09-29 on node 4 (ESP32-S3,
16 MB, rollback bootloader, OTA key provisioned). The
build from this branch was first written over USB. The pushes were made from
a Pi 5 with the `sensor-ota-push` cog.
1. Positive (MEASURED, S3). An image built from this branch was pushed to `ota_1`: HTTP 200,
1,230,848 B in 22 s.
- The node logged "httpd stack after POST /ota: 4900 of 8192 bytes never
used", so the upload's peak stack use is about 3.3 KB. With the old 1 KB
on-stack buffer that is over 4 KB, which is consistent with the
overflow.
- It rebooted and logged "OTA image pending verify on ota_1".
- `/ota/status` then read `ota_state: valid`.
- After a hard reset it came back on the same build in `ota_1`, still
`valid`, so the update is durable.
2. Negative (MEASURED, S3 forced-fail rollback). A `CONFIG_OTA_HEALTH_FORCE_FAIL=y` build was pushed to `ota_0`:
HTTP 200.
- `/ota/status` showed it running as `pending_verify` for about 30 s.
- The node then rebooted and came back on the previous build in `ota_1`,
`valid`, so rollback works.
3. One push attempt failed mid-upload when a serial capture was opened on the
node's USB-Serial-JTAG port at the same moment. The node reset and was
unharmed. Don't hold the USB console open while pushing OTA to a
USB-attached node.
- C6 positive, MEASURED on 2026-09-29 on node 5 (ESP32-C6, 8 MB
flash, 4 MB layout, rollback bootloader).
- The branch build was first written over USB and the OTA key provisioned.
- A second build was pushed from the Pi 5 cog to `ota_1`: HTTP 200 in 11 s.
- `/ota/status` showed `pending_verify`, then `valid` about 28 s later.
- After a hard reset it stayed on the new build, `valid`.
- Not yet verified, so CLAIMED (no log supports these):
- the 120 s no-IP timeout path (CLAIMED);
- a true power cycle (CLAIMED; a hard reset via RTS was used instead);
- the C6 negative/rollback test (CLAIMED; run on the S3 only).
- A pass of this check means only that the node reached the network and sent
CSI. It does not validate sensing quality, calibration, or anything else the
image does.
+58
View File
@@ -0,0 +1,58 @@
# ADR 380: Loopback UDP tee for second consumers of the CSI stream
## Status
Proposed. Unit tested, and smoke tested on loopback: a datagram sent to the
server arrived byte for byte at the tee target. Not yet exercised against a
live multi node fleet.
Date: 2026-09-28
Numbering: On 2026-10-02 `main` had up to ADR 378 and open PRs claimed nothing
higher; ADR 379 is the OTA first-boot health check. If another branch also
takes 380, renumber whichever merges second.
## Context
A five node deployment (four ESP32 S3, one ESP32 C6) streams ADR 018 frames to
one sensing server. A second consumer cannot see the raw stream.
Each node sends to a single `target_ip` (`firmware/esp32-csi-node/main/stream_sender.c`).
The server's WebSocket and REST outputs carry per node amplitude only: no
phase, no I/Q, no sequence number, and no `0xC511A110` sync packets. A fusion
host that aligns CSI with camera, radar, or lidar needs all of those. Binding a
second listener to the CSI port is not an option, and an external relay in
front of the port would make every source loopback, which the ADR 296
allowlist always admits.
## Decision
`--udp-tee` (env `RUVIEW_UDP_TEE`) copies every datagram the source allowlist
admits, byte for byte, to up to four `ip:port` targets.
- Targets must be loopback, so raw CSI never leaves the host through the tee.
- A target may not use port 0 or the listener's own port, which would loop.
- Sends are non blocking. A copy that cannot be queued is dropped and counted,
so a slow consumer never stalls the receive loop.
- The copy happens after the ADR 296 allowlist check and before any parsing,
so consumers see exactly what the server admitted, in every wire format.
- The tee is off unless configured.
## Consequences
A fusion host or Cognitum cog can subscribe to the full data plane by reading
a loopback port. A consumer that today binds the CSI port itself can keep its
code and move to the tee port instead.
The tee is byte transparent, so it needs no change when the ADR 018 header or
a later wire version changes.
Each target costs one `send_to` per admitted datagram. Four targets is the
limit.
## Validation
```bash
cd v2
cargo test -p wifi-densepose-sensing-server --no-default-features --lib udp_tee
```
+272
View File
@@ -0,0 +1,272 @@
# ADR-381: Node self-survey from pairwise ranging (check mode first)
- **Status**: Proposed
- **Date**: 2026-10-04
- **Deciders**: RuView maintainers; acceptance pending
- **Tags**: node-positions, ranging, ftm, uwb, multistatic, least-authority
- **Related**: ADR-018 (ESP32 wire format), ADR-110 (ESP-NOW time sync),
ADR-141 (privacy control plane), ADR-144 (UWB range-constraint fusion),
ADR-281 (BLE Channel Sounding ranging evidence), ADR-308 (placement
optimizer), rufield ADR-261 (BLE evidence and Channel Sounding adapter
boundary, `vendor/rufield/docs/ADR-261-ble-evidence-and-channel-sounding.md`)
- **Issues and PRs**: #1804, #1866, #2109 (show which nodes use the default
position), #2133 (give the governed fuser the configured node positions)
## Context
Node positions reach the sensing server only through `--node-positions`. The
operator types it by hand, keyed by node id (`id:x,y,z;...`) or by list index
when the prefix is omitted. Three kinds of mistake pass silently today:
- **Wrong ids.** A config keyed by ids that do not match the live nodes looks
applied and does nothing (#1804, #1866). #2109 makes the fallback to the
default position visible, and #2133 makes sure both fusers receive the
map. Neither can tell whether a correctly-keyed position is *right*.
- **Wrong units.** Positions measured in feet and typed as metres are 3.28×
too large. Nothing checks the scale.
- **Stale positions.** A node moved after setup keeps its old coordinates.
Each one degrades multistatic fusion without an error. The repository has no
node-to-node ranging and no anchor-free solver. ADR-144 models
anchor-to-*tag* UWB ranges for people tracking
(`wifi-densepose-mat/src/localization/range_constraint.rs`), not ranges
between the fixed nodes themselves. ADR-281 computes BLE Channel Sounding
ranging evidence, but there is no CS-capable radio in the fleet: rufield
ADR-261 notes that the ESP32-S3 does not implement CS.
The ESP32-S3 and C6 nodes do support Wi-Fi Fine Timing Measurement (802.11mc
RTT, station initiator ↔ SoftAP responder) in mainline ESP-IDF. So a ranging
source exists on hardware we already have. The question is how good it is.
### What FTM measured
Bench, 2026-10-04 (UTC): two ESP32-S3 boards (rev v0.2), ESP-IDF v5.5
`examples/wifi/ftm`, SoftAP responder on channel 1 at 20 MHz, 32 FTM frames
per session, line of sight, tape ground truth. All rows MEASURED; the
reproducer is in Appendix A.
| Tape (m) | Result |
|---|---|
| 0.00 | median 0.00 m |
| 0.76 | median 0.00 m. Readings clamp to 0 below about 0.8 m. The first three sessions after a responder start read 1.8 to 3.0 m |
| 1.14 | five responder starts: medians 0.60, 1.50, 2.25 and 3.08 m, plus a 7.7 m run at −68 dBm with a 1.92 m session spread (placement outlier) |
| 3.71 | median 7.80 m (one start) |
- **Precision within one responder start:** ±0.2 to 0.4 m (standard
deviation 0.19 to 0.31 m). A 40-session run over about two minutes showed
no drift beyond that.
- **Bias between responder starts:** −0.5 to +1.9 m at 1.14 m, with the
boards untouched. A plain-versus-auto-start responder A/B test showed the
responder build is not the cause.
- **Quantisation:** every distance is a multiple of 0.15 m (1 ns of RTT).
- **Not tested:** 40 MHz, other channels, `ftm -R -o` offset calibration, a
second pair, non-line-of-sight, CSI coexistence, raw t1..t4 timestamps.
FTM is precise inside one start and inaccurate across starts. It can catch
mistakes worth metres. It cannot place a node to decimetres. UWB (Qorvo
DW3000 family) is the expected accuracy path: about 0.1 m line of sight is
CLAIMED by the vendor, and we have no measurement of our own yet.
## Decision
### 1. A source-agnostic `RangeObservation` record
```text
RangeObservation {
initiator: u8, // node_id, as in --node-positions
responder: u8,
range_m: f64, // session median
sigma_m: f64, // session spread, 0 if unknown
method: Ftm | Uwb | BleCs | Manual,
samples: u16, // valid frames / exchanges
rssi_dbm: Option<i16>,
bandwidth_mhz: Option<u16>,
at_us: u64, // host receive time
responder_boot: Option<u32>, // boot counter or nonce of the responder
calibrated: bool, // set only by the host calibration step
}
```
`responder_boot` is required for calibration. The FTM bias moved with every
responder start, so an offset is only valid within the boot it was measured
in.
On the wire, nodes report ranges either in a new ADR-018-style packet with
its own magic, or as an extension of the existing sync packet. This ADR does
not choose between them. The host-side record and solver come first. The
wire format follows the first firmware that emits ranges.
### 2. Host-side check mode (warn-only)
Given the configured positions and a batch of observations, check mode
reports:
- **Per-link residuals.** `measured − ‖p_a − p_b‖`, with a tolerance of
`floor + k·σ`. σ is the larger of the median session sigma and the spread
across sessions. Each link is `ok`, `outlier`, `too_short` (configured
length below the source's minimum), `weak` (every session below the RSSI
gate), or `noisy` (σ above the source's maximum).
- **Unit mismatch.** A least-squares scale `s` in `measured ≈ s·expected`.
If `s` sits within a band of 0.3048 (feet), 0.0254 (inches), 0.01 (cm) or
0.001 (mm), and rescaling removes outliers, the check reports the unit.
- **Id swap.** For up to 8 ranged nodes, it scores every assignment of the
configured positions to ids. The cost is a truncated quadratic in residual
over tolerance, so one bad link cannot dominate. If another assignment
cuts the cost to at most half and the configured cost is material, the
check suggests that assignment, reported as `(node, position_of)` pairs.
- **Moved node.** If most of a node's links are outliers (at least 2, and a
strict majority), the node was probably moved. Outlier links left over are
reported one at a time as possible bias, a body or wall in the path, or a
bad session, and not as a configuration error.
- **Unknown and unranged nodes.** Observations naming unconfigured ids, and
configured nodes that no scored link reaches.
Distances do not change under rotation, reflection or translation, so check
mode compares distances and never coordinates. Where the layout is
symmetric, it says so instead of guessing:
- A relabelling that is a symmetry of the layout is **invisible**. For
example, ids rotated 180° in a rectangle of corner nodes. Check mode
reports such a layout as consistent.
- When several assignments fit equally well, check mode prefers the one that
changes the fewest ids, since a typo moves few. It reports how many
assignments tied. If the fewest-changes assignments are themselves tied,
as with a swap along one wall of a rectangle (its mirror image is the
opposite wall's swap), it reports `id_swap_ambiguous` and names nothing.
Corner-mounted nodes in a rectangular room are the common case and are
nearly symmetric. Operators should expect ambiguity reports there, and a
fifth node off the axes removes most of them.
**Authority.** Check mode never writes positions, never edits config, and
never changes fusion. Its output is a report for the startup log and a
read-only status field. The report shape is JSON-serialisable; the field
name follows #2109's `node_positions` block once that lands.
**Tolerances come from measured data:**
| Source | `floor_m` | `calibrated_floor_m` | `max_sigma_m` | `min_expected_m` | Basis |
|---|---|---|---|---|---|
| FTM, ESP32-S3, 20 MHz | 2.0 | 0.6 | 1.0 | 1.0 | MEASURED, bench above: bias −0.5 to +1.9 m between starts; spread ≤ 0.31 m within a start (the 1.92 m outlier excepted); clamp below about 0.8 m |
| UWB | 0.3 | 0.3 | 0.3 | 0.0 | ESTIMATE until our own UWB bench. Vendor accuracy is CLAIMED |
With `floor_m = 2.0`, the one 3.71 m start (+4.1 m) would be flagged as a
single-link outlier. That is uncalibrated FTM as it stands. Single-link
outliers stay low-severity on purpose, and the strong findings (unit, swap,
moved node) need several links to agree.
### 3. Per-boot calibration
An offset is recorded per `(initiator, responder, responder_boot)` from
sessions at a known reference distance of at least 1.0 m:
`offset = median(range) − reference`. It applies only to observations with
the same key. After a responder reboot, observations pass through
uncalibrated and get the wider floor.
Zero-separation calibration (ESP-IDF's `cm0` / `ftm -R -o`) is not used,
because short ranges clamp to 0 on this hardware. In practice, the reference
is one tape-measured link per responder per boot, or a UWB link once UWB
exists. That the bias is stable over a whole boot, and not only for two
minutes, is untested (ESTIMATE). Measuring it is the first FTM follow-up.
### 4. Fill mode, gated off
Fill mode (solve positions with classical-MDS initialisation, weighted
robust SMACOF and gauge fixing, then write a file the operator passes to
`--node-positions`) is **not implemented** and stays disabled for a source
until all of these hold:
- the source has **MEASURED** link-error evidence with a reproducer;
- the evidence covers at least 6 ground-truth links;
- the p90 link error is **≤ 0.3 m**.
FTM's measured errors are metres, so it fails this gate. UWB is expected to
pass it, but that is unproven. Even when the gate opens, fill mode writes a
proposal file. The server never rewrites its own positions.
### 5. Pluggable sources
Every source produces `RangeObservation`, and check mode does not care which
one did. Only the tolerance profile changes.
| Source | Status | How it plugs in |
|---|---|---|
| **FTM** (ESP32-S3/C6) | Now, for check mode | APSTA node runs a short scheduled survey window. One node is the SoftAP responder and the others initiate a directed session. `rtt_ns`, RSSI and session spread map to `range_m`, `rssi_dbm`, `sigma_m`. Airtime cost on the CSI channel is unmeasured |
| **UWB** (DW3000 family, e.g. DWM3001CDK over USB) | After its own bench | Two-way ranging between nodes. Shares hardware with ADR-144's anchor-tag model, but node↔node ranges feed this ADR and anchor↔tag ranges keep feeding ADR-144 |
| **BLE CS** | When a CS-capable radio exists | Through the rufield ADR-261 adapter boundary, using ADR-281's cross-validated phase/RTT ranging. A `Divergent` anomaly drops the observation. Single-mechanism evidence gets the wider floor |
| **Manual** | Now | A taped distance entered by the operator. Mainly useful as the per-boot calibration reference |
### 6. Prototype
`v2/crates/ruview-survey` implements sections 1 to 4 on the host, with no
I/O, no clock and no randomness, and bounded inputs: 64 nodes, 10 000
observations, 100 m range, and the swap search capped at 8 nodes.
Integration with the sensing server is a follow-up and depends on #2109 for
where the report surfaces. The crate's tests are SYNTHETIC: generated layouts
with FTM-like bias and noise from the envelope above. They cover consistent
configs, feet and centimetre mistakes, a two-id swap, a three-id rotation,
symmetric invisibility and ambiguity, a moved node, a bench-style single-link
outlier, a noisy link, per-boot calibration, and the fill gate.
`examples/check_ranges.rs` runs check mode on a JSONL file.
## Consequences
- An operator gets a warning when ranges and `--node-positions` disagree
grossly, with no new authority: check mode is read-only.
- FTM on nodes needs APSTA and a SoftAP beacon on the sensing channel. Its
cost to CSI yield must be measured before the firmware side ships.
- The FTM tolerances rest on one evening with one pair. They are parameters,
not constants, and must be revised from more data (40 MHz, more pairs,
non-line-of-sight).
- Ranges between fixed nodes are not person data. Ranges to phones or tags
are, and remain under ADR-141 and ADR-144.
- A symmetric layout can hide a relabelling entirely. Check mode documents
this and does not claim to catch it.
## Evidence rule
No self-survey accuracy claim without a MEASURED reproducer: tape or laser
ground truth, at least 4 nodes, line of sight plus at least one
non-line-of-sight case, every number tagged. The prototype's tests are
SYNTHETIC and show logic, not radio performance.
## Alternatives considered
- **Treat FTM as accurate enough to fill positions.** Rejected: the measured
bias between starts (−0.5 to +1.9 m at 1.14 m) and the +4.1 m at 3.71 m
rule it out.
- **Drop FTM.** Rejected for now. It costs nothing on existing hardware, it
catches metre-scale mistakes, and it may suit coarse gating, calibrated
pairs within a boot, and time sync. It is parked and revisited after UWB.
- **Zero-separation `cm0` calibration.** Rejected: readings clamp to 0 below
about 0.8 m.
- **Server auto-corrects positions.** Rejected under least authority. Fill
mode, when enabled, writes a proposal file only.
- **LoRa SX1280 ranging, router (MT7981) FTM, raw I/Q phase ranging,
CSI-amplitude ranging.** Rejected or deferred: wrong distance regime, no
driver support, unproven, or too inaccurate respectively.
## Open questions
- Does one calibration offset hold for a whole responder boot? For both
roles? For other initiators against the same responder?
- Is the post-start transient (1.8 to 3.0 m in the first three sessions at
0.76 m) real, and how long does it last?
- Does 40 MHz shrink the bias between starts?
- What does an FTM survey window cost in CSI frames on the same channel?
- Can FTM t1..t4 timestamps be tied to the CSI timestamp clock for sync?
- The wire format: new magic, or an extension of the sync packet?
## Appendix A: FTM bench reproducer
1. Flash ESP-IDF v5.5 `examples/wifi/ftm` on two ESP32-S3 boards
(`CONFIG_ESP_WIFI_FTM_ENABLE=y`, the example's default).
2. Responder console: `ap <ssid> <password>` (channel 1, 20 MHz defaults).
Use throwaway credentials and do not commit them.
3. Initiator console, at a taped separation: `ftm -I -s <ssid>`, 32 frames,
10 to 40 sessions per distance. Record `Avg raw RTT`, `Avg RSSI`,
`Estimated Distance` and the valid-frame count for each session.
4. Reset the responder and repeat at the same distance to see the bias move
between starts.
5. Report the median, spread and error per run, tagged MEASURED, with the
raw per-session data alongside.
@@ -0,0 +1,137 @@
# ADR-382: `spatial.evidence.v1` export of RF Gaussians and link observations
- **Status**: proposed
- **Date**: 2026-10-03
- **Deciders**: RuView maintainers
- **Tags**: rf, gaussian, spatial, evidence, wire-format, interop
- **Related**: ADR-273 (unified RF world model), ADR-275 (RF-aware Gaussian
memory), ADR-142 (temporal voxel evidence). Origin of the wire format:
WeftOS ADR-107 §7 ("Spatial evidence engine and workspace"), amended
2026-10-03.
- **Crate**: `v2/crates/ruview-spatial-evidence`
Numbering: on 2026-10-03 `main` had up to ADR-378, open PRs claimed 379
and 380, and 381 was taken by another branch.
## Context
`ruview-unified` keeps a map of RF Gaussians (ADR-275) and a Friis channel
model that turns a TX/RX link into an expected amplitude. A spatial evidence
engine fuses many sensor types (tape measurements, radar, ToF, UWB, RF) into
an occupancy octree and promotes stable geometry (walls, furniture) into room
models. That engine reads a versioned JSONL format, `spatial.evidence.v1`,
specified in WeftOS ADR-107 §7.
RuView should be able to feed it without depending on it, and without a
reader of this repository having to go to another project to learn what the
export writes. This ADR records the part of the contract RuView emits.
## Decision
1. A leaf crate, `ruview-spatial-evidence`, converts `ruview-unified` output
to `spatial.evidence.v1` lines. Its only internal dependency is
`ruview-unified`. The wire shapes are local serde structs; the contract is
the JSON, not a shared crate.
2. It emits two record types, `rf_gaussian` and `rf_link_observation`. It
does not mirror the other v1 types and rejects them as unknown when
parsing.
3. Every record is validated against the v1 rules below before it is
serialised, so a bad value fails in RuView rather than in the consumer.
4. If WeftOS ADR-107 §7 changes the envelope or either RF body, this ADR is
amended in the same change that updates the crate.
### Envelope (every record)
| Field | Rule |
|---|---|
| `schema` | exactly `spatial.evidence.v1` |
| `type` | `rf_gaussian` or `rf_link_observation` here |
| `t_ns` | u64, ns since the Unix epoch |
| `frame` | `room_enu` |
| `region` | region id, e.g. `region/urth/meso/<room>` |
| `source_id` | device or operator id, never a person |
| `uncertainty_m` | 1σ position uncertainty, metres, in (0, 100] |
| `provenance` | `{receipt, producer, proof}`; `proof` is `MEASURED`, `CODE` or `SYNTHETIC` |
Ids (`region`, `source_id`, `receipt`, `producer`) match
`[A-Za-z0-9._:/-@]{1,128}`. A line is at most 16 KiB.
**Frame.** `room_enu` is room-local, origin at the room's SW floor corner,
x east, y north, z up, metres. Coordinates are within ±1000 m.
**Angles.** Every yaw in v1 is degrees counter-clockwise from room +x
(east): 0 faces +x, 90 faces +y. The two RF types have no yaw field. The
`rf_gaussian` orientation is a body-to-room quaternion, and a positive
rotation about +z turns the Gaussian's local x axis counter-clockwise from
east, the same sense.
### `rf_gaussian`
| Field | Rule |
|---|---|
| `position` | [x, y, z] m |
| `scale` | per-axis σ, m, each in [1e-6, 1e4] |
| `orientation` | quaternion `[w, x, y, z]`, finite, norm > 1e-6 |
| `occupancy` | peak extinction, nepers/m, in [0, 1e6] |
| `confidence` | [0, 1] |
| `motion` | `static`, `slow` or `fast` |
| `role`? | `absorber` or `reflector`; set only on priors sent back to RuView |
The exporter maps `RfGaussian` field for field. `uncertainty_m` is the
geometric mean of the three σ, clamped to the envelope range, because
`RfGaussian` has no separate position covariance. The receipt is
`gauss:<device>:<t_ns>:<index>`. A Gaussian whose provenance is `synthetic`
is always tagged `SYNTHETIC`; a caller can lower the proof tag, never raise
it.
### `rf_link_observation`
| Field | Rule |
|---|---|
| `tx`, `rx` | antenna positions, m |
| `freq_hz` | [1e8, 3e11] |
| `excess_loss_db` | `20·log10(free / measured)`, in [−60, 200]; negative is constructive multipath |
The free-space reference comes from `ruview-unified`'s channel model with an
empty map, so it is the same Friis amplitude the ADR-275 inverse update uses.
### Examples
These two lines are the WeftOS ADR-107 §7 examples. The crate's tests parse
them and check that its own output has the same shape.
```jsonl
{"schema":"spatial.evidence.v1","type":"rf_link_observation","t_ns":1759500001300000000,"frame":"room_enu","region":"region/urth/meso/test-room","source_id":"node-2","uncertainty_m":1.0,"provenance":{"receipt":"csi:node1-node2:000881","producer":"ruview-adapter@0.1","proof":"MEASURED"},"tx":[6.20,3.50,1.10],"rx":[0.20,0.20,1.10],"freq_hz":2437000000.0,"excess_loss_db":4.2}
{"schema":"spatial.evidence.v1","type":"rf_gaussian","t_ns":1759500002000000000,"frame":"room_enu","region":"region/urth/meso/test-room","source_id":"ruview-unified","uncertainty_m":0.5,"provenance":{"receipt":"gauss:7f3a","producer":"ruview-unified@0.3","proof":"CODE"},"position":[3.2,0.0,1.2],"scale":[0.05,1.6,1.2],"orientation":[1,0,0,0],"occupancy":0.7,"confidence":0.8,"motion":"static","role":"absorber"}
```
### Versioning
Additive optional fields may appear inside v1 and older readers ignore them.
New record types may also be added inside v1 (the 2026-10-03 amendment added
`tof_depth` and `radar_range`, and an optional `sensor` block on
`radar_track_point`); a reader that does not know a type rejects it as
unknown rather than misreading it. Removing, renaming or changing the
meaning of a field requires `spatial.evidence.v2`, and the parser reports a
v2 line as a version mismatch.
## Consequences
- RuView output can feed a spatial evidence engine with no new runtime
dependency, and the contract is readable from this repository.
- The engine can send promoted walls and objects back as `rf_gaussian`
priors with a `role`. That gives ADR-275's inverse update a geometric
starting point instead of an empty map.
- The struct definitions are duplicated between the two projects. The
golden-line tests catch drift in the RF types; a change on the WeftOS side
still needs someone to amend this ADR and the crate.
## Limitations
- Nothing calls the crate yet. The sensing server does not export records,
and there is no adapter that turns an inbound `rf_gaussian` prior into an
`RfGaussian` in the live map.
- The tests are CODE-level (deterministic conversions and validation). No
MEASURED claim is made about what the exported evidence does to a room
model.
- `uncertainty_m` for a Gaussian is a stand-in derived from its extent.
@@ -0,0 +1,23 @@
# ADR-383: Reject malformed HTTP request targets without stopping the transport
Status: Proposed; maintainer acceptance pending
## Context
ADR-375 introduces the authenticated Streamable HTTP transport. Node accepts an HTTP request target that the WHATWG URL parser rejects. The current request listener lets that parser exception escape, terminating the process before route selection. A bounded loopback fixture using the public `startMcpHttp` API reproduces this with `http://[`; the original code resets the connection and exits.
## Decision
Catch URL construction failures at the request boundary and return HTTP 400 with the generic `invalid_request_target` error. Do not include the request target or parser exception in the response, since the target can carry a secret path. Keep route selection, token checks, origin policy, grants, protocol negotiation and JSON handling unchanged.
## Alternatives
Relying on Node's HTTP parser does not suffice: a syntactically accepted request target can still fail URL construction. A process-level exception handler would obscure unrelated faults and would not provide a request-specific response.
## Verification
MEASURED software fixture: the actual child-process HTTP transport fails the new regression on the original source (nine existing controls pass); the patched transport passes all ten focused tests. The new test also verifies subsequent health, rejected unauthenticated calls and successful authenticated initialization. Broader package checks are recorded separately against the final implementation revision. No external service, device, hardware, throughput or production deployment claim is made.
## Consequences
Malformed URL targets receive a stable client error and do not interrupt later requests. This change does not add SSE or sessions, change authentication, or authorize additional operations.
@@ -0,0 +1,19 @@
# ADR-384: Preserve explicitly supplied SDK search limits
Status: Proposed; maintainer acceptance pending
## Context
The public `memorySearch(query, limit)` helper uses a truthiness check before forwarding its optional limit. Explicit zero and NaN therefore disappear, even though the shared registry rejects them. The helper silently returns success where the equivalent raw SDK call returns `invalid_arguments`; strict SDK users also miss the expected `RuViewError`.
## Decision
Omit the optional limit only when it is undefined. Forward every explicitly supplied value to the existing shared validator. Do not duplicate validation or change the registry range, search algorithm, result count, strict error policy or defaults.
## Alternatives
Validating separately in the helper would duplicate schema rules and risk drift. Retaining truthiness makes invalid caller inputs indistinguishable from an omitted option.
## Verification
MEASURED local software regression exercises the public SDK and actual registry. Original code fails for zero; the corrected helper rejects zero, NaN, negative and excessive limits, preserves strict errors, and matches raw registry results for omitted and valid limits. Package gate receipts are recorded separately against the final revision. No external retrieval, training, hardware or production service claim is made.
@@ -0,0 +1,19 @@
# ADR-385: Stop watch mode on permanent input errors in every output mode
Status: Proposed; maintainer acceptance pending
## Context
The CLI's `esp32 --watch` repeats short capture windows. Permanent bind/port errors stop the terminal view, but the piped/JSON branch continues before reaching that check. Schema-level invalid arguments also never stop either branch. Actual CLI children with a fractional UDP port or rejected bind value repeatedly emit errors despite having no usable capture configuration. These fixtures fail before opening a socket or accessing a device.
## Decision
Classify existing permanent port/bind errors and schema-invalid arguments before selecting output mode. Emit the error once and return exit status 1 for both terminal and JSON output. Keep repeated capture behavior for nonterminal results, including windows with no CSI packets; retain the normal watch output contract.
## Alternatives
Adding a delay would still repeatedly retry an unchanged invalid configuration. Stopping every unsuccessful capture would incorrectly terminate legitimate empty capture windows.
## Verification
MEASURED bounded actual CLI regression checks piped `--watch --json` with invalid inputs, one structured failure and nonzero exit. No network socket, hardware, external provider or live capture is exercised. Whole-package validation receipts accompany the final revision.
@@ -0,0 +1,23 @@
# ADR-386: Recover CSI packet loss by missing sequence identity
Status: Proposed; maintainer acceptance pending
## Context
The RF capture diagnostics report CSI packet loss and account for UDP reordering. The existing tracker decrements its gap count for any late sequence while gaps exist. A duplicated recovery or previously received packet therefore erases a different missing packet. For example, sequences 100, 103, 101, 101 leave 102 missing but currently report zero loss. Accurate loss diagnostics matter when distinguishing transport problems from healthy sensing input.
## Decision
Retain the identities of counted missing sequences within the existing 1024-sequence reorder window. Recover a gap only when deleting its identity succeeds, so a duplicate cannot recover it twice. Prune identities outside the window without erasing their historical gap counts. A confirmed counter reset clears recovery identities but preserves accumulated loss. Keep the existing frame count/rate, loss denominator, stray/reset classification and packet format unchanged.
## Alternatives
A duplicate-only set cannot distinguish packets observed before the capture baseline from genuinely missing ones. Keeping all missing identities indefinitely is unnecessary and could grow without bound.
## Verification
MEASURED software fixtures use byte-accurate RAC1 envelopes, the public capture API and real loopback UDP. Regressions cover duplicated recoveries, already-received packets, pre-capture packets and distinct valid recoveries. A direct public parser/NodeStats control checks bounded bookkeeping and reset semantics. Existing sequence/stray/reset tests remain applicable. This is a software diagnostic correction, not physical RF, device throughput or physiological accuracy validation.
## Consequences
Loss no longer disappears when unrelated late packets arrive. Recovery remains limited to the established reorder window; older gaps stay counted. Stored identities are bounded per node.
@@ -0,0 +1,32 @@
# ADR-389: Measure CSI rates over the observed capture interval
- Status: proposed
- Date: 2026-10-11
- Related: ADR-373 (receive-only device diagnostics)
## Context
The public `captureEsp32` producer may stop at `max_packets` before its configured
`seconds` deadline. Its node CSI rates still divided received packets by that
requested deadline. Two packets received in approximately 150 milliseconds with a
30-second deadline therefore appeared as 0.07 Hz, misleading operators evaluating
whether a node streams normally.
## Decision
Start a monotonic clock when the UDP socket successfully binds. At completion,
calculate `elapsedSeconds` and use it as the denominator for node CSI rates.
Retain `seconds` as the configured deadline for compatibility and expose the
observed interval separately. Close and analysis processing time are excluded.
A socket error before binding has a zero observed interval and no received data.
## Verification and limits
A deterministic-format ESP32 fixture is transmitted through an owned loopback UDP
socket into the public capture function. The baseline fails because the observed
interval is absent and the rate is 0.07 Hz. The fixed capture checks early
termination, preserves the deadline, and checks the rate against observed elapsed
time. Existing CSI decoding and capture tests remain controls. This is software
transport verification; it makes no radio, hardware, sensing-accuracy, or firmware
throughput claim. Normal deadline captures may report a slightly longer interval
because timer callbacks are scheduled by the host event loop.
@@ -0,0 +1,27 @@
# ADR-390: Reduce LiDAR scan extrema without variadic argument limits
- Status: proposed
- Date: 2026-10-11
- Related: ADR-373 (device diagnostics)
## Context
`readRplidar` accepts up to 4 MiB of serial scan bytes. Its summary passed every
valid distance as an argument to `Math.min` and `Math.max`. A valid 1 MiB scan
containing 200,000 measurement nodes exceeds the JavaScript argument-stack limit
and throws `RangeError` instead of returning diagnostics.
## Decision
Reduce extrema with a loop and constant auxiliary state. Retain current median,
zero-distance filtering, coverage, units, and empty-scan behavior. This avoids
runtime-specific argument limits across the full supported capture size.
## Verification and limits
An offline serial-pump dependency supplies protocol-valid bytes to the public
`readRplidar` producer. The 3-point control passes on the baseline; the 200,000
point fixture fails with `RangeError`. Both pass after this change with exact
minimum, median, maximum, and coverage checks. Existing device tests provide
empty and mixed-input controls. The fixture uses no serial device or motor and
proves software decoding/aggregation only, not real hardware performance.
@@ -0,0 +1,29 @@
# ADR-391: Retain synthetic CSI provenance in analysis input
- Status: proposed
- Date: 2026-10-11
- Related: ADR-323 (RAC1 synthetic flag), ADR-373 (CSI analysis)
## Context
The RAC1 decoder exposes the firmware's synthetic flag and the waterfall retains
it, but the kernel-analysis frame collector discarded it. Analysis of synthetic
input consequently returned the same live-CSI explanatory note as unmarked input,
preventing callers from distinguishing synthetic-inclusive estimates.
## Decision
Track whether any retained analysis frame is firmware-marked synthetic, expose
that boolean in `analysis.input.synthetic`, and explicitly describe synthetic
input in the estimate note. Frames rejected by the analysis buffer cap do not
change the provenance of the retained computation. Keep arrival-rate measurements
and raw packet diagnostics unchanged. An unmarked packet is not proof of physical
radio origin; this records an explicit upstream flag rather than certifying truth.
## Verification
The public capture producer receives protocol-valid RAC1 datagrams through owned
loopback UDP and invokes an injected deterministic analyzer. A mixed marked and
unmarked window fails the baseline provenance assertion. The fixed path retains
true for the mixed window and false for an unmarked control. This validates
software provenance propagation only, not sensing accuracy or hardware behavior.
@@ -0,0 +1,27 @@
# ADR-392: Return single valid frames for explicitly selected radar models
- Status: proposed
- Date: 2026-10-11
- Related: ADR-373 (host device diagnostics)
## Context
The serial radar reader requires two valid frames before auto-detecting a model.
That confidence threshold was also applied to an explicitly selected model,
rejecting a valid short capture with one decoded presence or vital report as
`no_valid_frames`. The caller already supplied the model and baud selection.
## Decision
Require two frames for auto detection and one for an explicitly selected model.
Retain the empty-capture failure, checksums, parsing, capture permissions, and
all reported values. A single decoded frame is a successful software read and
is not evidence of clinical accuracy or sustained radar performance.
## Verification
An offline serial-pump fixture supplies one checksum-valid MR60BHA2 presence
report through the public reader. Explicit model selection fails on the baseline
and returns one frame with presentFraction=1 after the change. The same input in
auto mode remains refused. Existing multi-frame and invalid-input device tests
remain controls. No serial port or hardware was opened by this verification.
@@ -0,0 +1,28 @@
# ADR-393: Use the arithmetic median for even LiDAR samples
- Status: proposed
- Date: 2026-10-11
- Related: ADR-373 (range and depth diagnostics)
## Context
The shared LiDAR median helper selected the upper middle sample for even-sized
sets. The public depth reader reported 4 metres for confident depths of 1 and
4 metres, and the serial scan reader likewise returned 4000 mm. Fields named
median should report the arithmetic median, 2.5 metres or 2500 mm respectively.
## Decision
Average the two central sorted samples for even counts and preserve the central
sample for odd counts. Empty sets remain null. Existing distance/confidence
filtering and metre/millimetre conversion remain in the public producers.
Update existing even-sample expectations to reflect the corrected numerical
contract; this intentionally changes these statistics for even counts.
## Verification
Offline protocol fixtures enter both public readers through their existing
transport injection boundaries. The baseline fails two even-sample cases and
passes an unsorted odd-sample control. The correction verifies both units and
zero-depth exclusion. No real LiDAR, serial device, or radar was used, and this
software calculation does not prove measurement accuracy.
@@ -0,0 +1,26 @@
# ADR-394: Apply strict SDK errors to diagnostic reports
- Status: proposed
- Date: 2026-10-11
- Related: ADR-369 (SDK operator interface)
## Context
A strict SDK instance throws `RuViewError` when an operation returns `ok:false`.
The doctor helper directly invokes the diagnostic producer to retain injectable
operator dependencies, bypassing that shared result check. Failed diagnostics
therefore silently resolved even when the caller selected strict mode.
## Decision
Extract the existing result check and apply it to both registry calls and doctor
reports. Preserve injected dependencies and retain the original structured report
on the thrown error. Non-strict calls and successful diagnostics continue to
resolve reports; the existing dry-run exception remains unchanged.
## Verification
The public SDK doctor runs a deterministic offline invalid-protocol diagnostic.
The baseline fails the promised strict rejection. The correction checks the exact
error class, tool identifier, and failed report; controls verify non-strict failure
reports and successful strict reports. No external service or device is accessed.
@@ -0,0 +1,27 @@
# ADR-395: Preserve missing numeric CLI values for schema rejection
- Status: proposed
- Date: 2026-10-11
- Related: ADR-369 (operator CLI)
## Context
The common device/training CLI flag adapter coerces values with `Number`. A bare
`--node-id` is parsed as boolean true and becomes node 1; `--node-id=` becomes
node 0. Both silently select a real node rather than reporting the missing
numeric argument. The same adapter handles capture deadlines and sample limits.
## Decision
Retain boolean and empty-string values in the common schema argument adapter so
existing numeric schemas reject them. Convert explicit non-empty numeric values
as before, including zero. Keep presentation flags and tool schemas unchanged.
This change is limited to verbs using the common snake-case argument adapter.
## Verification
Execute the actual CLI process with both missing-value forms and explicit zero.
An intentionally invalid UDP port ensures argument validation completes before
opening a socket. Baseline errors omit node_id for both missing forms; the fixed
CLI reports those numeric type errors while preserving the explicit zero control.
This is offline CLI/schema verification, not a radio capture.
Loaded 100 of 507 files, more files were not shown because too many files have changed in this diff. Show more