The page embedded all 21 languages' cw_* strings (174 KiB of the 220 KiB gzip
budget). Now it embeds English only; every other language is a separate
build artifact (dist-console/strings/<lang>.json) gzip'd into the helper and
served at GET /strings/<lang>.json (Host/Origin checked, no token, immutable
year-long cache; the page asks with ?v=<hash of the strings>). The page shows
English until the chosen language arrives, falls back to English per key, and
stays in English (logged once) when a load fails.
Page: 607.1 KiB raw / 174.4 KiB gzip -> 357.5 KiB raw / 112.5 KiB gzip.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Fix round 2. The helper refused a request under 500 ms after the last one (429),
which caught the final batch sent as Controls closes, right after a regular one,
so ○ was still lost. The gate is now a pure, host-tested function: one request
marked {"final":true} is admitted inside the window, once per window, each body
still at most 4 KiB. The page marks its final batch (keepalive, cut to fit with
the flag) and sends it only after a regular send still on the wire, so the two
never arrive out of order.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Fix round 1 for the controls guide. Controls shows "Works" only for input the page
acts on today (nav.ts's always-live keys, verified bindings, the pointer) and
"Received" for keys that arrive but are not used yet. Stopping the input log, or
the page being hidden while it runs, sends the queued events at once with fetch
keepalive, so ○ closing Controls is not lost. Legend labels wrap to two lines and
fit in all 21 languages; the drive list's Try again shows Loading; the focused
drawing keeps its ring without the washed-out fill; unused cw_settings_display
removed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
A Controls screen (Settings, the rail's Help, or OPTIONS once measured) draws the
DualSense with each button's job on the page and lights every button the page
receives ("Works"); PS and Mute are marked as used by the PS5, the rest "not
received yet". While it is open the input log is on, and keys do nothing else
(no click, no focus move, no browser reload) except ○, which closes it.
Three first-run tip cards (sections; ✕ selects, ○ goes back, everything has an
on-screen button; the PS5 pointer and full screen) show once, skip with ○ or
Skip, and come back from Settings › Show tips again. A help overlay lists the
buttons that work on the current screen and those not confirmed yet. The hint
bar shows ✕ and ○ plus only verified buttons with a job on the screen in front.
Settings order: Theme · Language · Full screen · Controls · Show tips again ·
Allow other devices · home screen tile · About. Empty and failed states say what
happens next (drive list and settings get Try again). CHANGELOG 7.1.0 describes
the final page. Gzip budget 220 KiB (ruling R19).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
The console page gets a Cheats section (rail: Home, Files, Cheats, Settings; Home's
Cheats shortcut now shows). It lists the running game's cheats as switches: X or the
square button switches the focused one, rows are inert while a switch runs, and the
helper's state is read again after every switch. No game, no cheats, the game closing
while the page is open, and helper errors each say what to do next, with Refresh.
The helper adapts the cheat methods exactly as the PC engine calls them:
/api/cheats/status (CHEATS_STATUS), /api/cheats/list {title_id} (CHEATS_GET, every
entry re-escaped, a cut reply keeps its whole entries) and /api/cheats/toggle
{title_id,index,on} (CHEATS_TOGGLE, its free-text refusals mapped to tokens).
Downloads and deletes stay in the PC app.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
- Hover never focuses a text field; a field holds focus against hover only for
1.5 s after a key typed into it, so the ring cannot stay on an input while
the pointer (and the native ✕ click) sits on OK.
- Ruling R18: only verified bindings act; the always-live set is the arrows,
Enter, Escape and Backspace (and the old gamepad buttons). F5, Home,
PageUp/PageDown, ContextMenu, F1, F11 keep their browser defaults. PageUp/
PageDown are left to text fields once verified.
- Full screen: a refusal (error event or rejected promise) hides it for the
session; a webkit request counts only when the page really went full
screen; automatic only in the PS5 browser; Escape spends no try; the
Settings switch follows fullscreenchange.
- cw_hint_filter removed from all 21 locales.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
The PS5 browser drives an on-screen pointer and ✕ clicks at it, so the page
now treats hover as focus: the ring and what ✕ activates are the same
element. After a key move the pointer is parked (drift under 6 px is
ignored) and a click there activates the focused control instead.
nav.ts's button table covers every DualSense button with its action, the
keys/keyCodes it may arrive as and verified:false until measured: L2/R2 page,
L3 top, R3 rail, OPTIONS help, Create refresh, touchpad full screen, PS/Mute
system. The hint bar shows ✕/○ always, the rest once verified.
POST /api/input-log appends compact, privacy-neutral event lines to a 64 KiB
ring at /data/ps5upload/webui-input.log (same-directory rewrite, bounded
body, 429 under 500 ms); the page sends only while setInputLogEnabled(true).
Full screen on the first activation (standard, then webkit-prefixed), with a
Settings switch hidden where unsupported. Type scales with the viewport
(26 px at 1540x700). Raw page budget 640 KiB.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
- Copy/move/delete batches send the next item only after the previous
item's job left "running" (files/batch.ts: page-wide batch, survives
section switches, "Copying 3 of 40", Stop ends it before its next item).
- The PS5 browser's ○ is history back: one history guard sits on top
while the page has somewhere to go back to (modal, folder, a section
other than Home); popstate runs the page's back action. Escape,
Backspace and a gamepad ○ call history.back() through the same path.
- Home's confirm dialog keeps 409/503 errors inside it; the status poll
pauses while the page is hidden; images over 16 MiB say "too large to
preview"; storage cards are role=button with Enter/Space; the tile
switch waits for the helper's real result (tile_busy/tile_failed).
- New cw_ strings in all 21 locales; three unused ones removed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
The rail and the content are nav regions ([data-nav-region]). Up/Down stay
in the focused region and stop at its ends. Left/Right move within the
region while something lies straight that way, else enter the neighbouring
region at its entry: [data-nav-entry] (the rail's active section), else the
control last focused there if it is still shown, else its first control.
Modals keep their own scoping. ✕ on a Home storage card opens Files at
that drive's path.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Closing the viewer aborts its reads (no more 1 MiB requests holding a
helper web worker); copying or moving a folder into itself or a subfolder
is refused on the page with its own message; a 409 says "already exists"
only for an exists code; opening Files at a path (Browse USB) resets the
selection and focuses the first row like any folder change.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Drives -> folders (folders first, natural name order, sizes, free space),
paged 256 at a time through /api/fs/list's next offset; "Show more" loads
the next page when the D-pad reaches it. The actions sheet (triangle or the
on-screen "..."): View, Copy to..., Move to..., Rename, New folder, Info,
Delete (confirm names the item and the count). Select (square or the
toggle) applies Copy/Move/Delete to several items, sent one request at a
time; Copy/Move/Delete are off while a batch is being started. Moves rename
on the same drive and fall back to a move copy job on 409 cross_mount; copy
and move send each item's full destination path, as JOB_COPY and the
engine use it. Viewer for png/jpg (blob URL, revoked on close) and
txt/log/json/ini/xml (first 1 MiB as text nodes, D-pad steps through it).
Circle goes up one folder or closes the open modal.
ConfirmDialog (and every new modal) restores focus to the row now at the
opener's place when the opener is gone; Files also refocuses that place
when a refresh removes the focused row. App's circle handler is now a
nav fallback so a section's own circle runs first. Home's Browse USB
appears, since Files opens at ctx.params.path.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Running game with Close, one storage card per drive, every job with its
owner (this console / a PC) and Cancel, quick actions Browse USB (hidden
until Files opens at a path), Rest mode and Restart. Every destructive
action goes through the shared ConfirmDialog (nav modal, focus on the safe
choice, ○ closes). useJobs polls /api/jobs every 1 s while a job runs,
every 5 s otherwise, stops on unmount and on 401. Sections get a context
(status, refreshStatus, openSection with params). Strings in 21 locales.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
- nav.ts: while a [data-nav-modal] is open only handlers scoped inside it
run (onNavAction's new scope getter), so the reload banner (and later
dialogs) keep ○/L1/R1 from moving focus to the page behind; confirm
with focus outside the modal brings focus in instead of clicking.
- R12: an Arrow keydown within 100 ms after the gamepad fired the same
direction is dropped as the browser's echo.
- GET / adds frame-ancestors 'none' and X-Frame-Options: DENY.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
The page the helper serves on the PS5 itself: client/src/console/ built by
vite.console.config.ts into one self-contained dist-console/index.html
(script and styles inlined, safari13, 253 KiB raw / 84.5 KiB gzip; the
build fails over 450 KiB raw or 150 KiB gzip). payload/Makefile gzips it
into build/webui_page.c and the helper serves it at GET /, with
GET /api/token for the per-start token.
- Shell: top bar (logo, console chip from /api/status), rail of the
implemented sections (Home, Files, Settings placeholders), hint bar
showing only verified bindings, reload banner on a 401.
- nav.ts: spatial focus (aligned neighbours first), the R11 default input
table (keyboard + Gamepad "standard", all unverified), D-pad repeat.
- api.ts: token fetched once, X-PS5U-Token on every POST, 401 → expired.
- strings.ts: cw_* keys from all 21 locales via build-console-strings.mjs
(vm-eval like the i18n gate), English fallback.
- CI: client npm ci before every make payload.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Per the 2026-10-16 rest-mode detection spec (client side). The engine's
status verdict (power: awake/rest/unreachable, reason console_rest_mode)
is read defensively, so this builds alone and lights up once the engine
side lands.
- HostRuntime.power, set by the status poller; rest flips the console down
at once (no miss debounce) and is logged on the edge.
- Status bar, console tabs, Home badge/metric, game page console rows:
"In rest mode" in the calm accent, "Not answering" for unreachable.
- ConnectionGate: rest-mode notice with Wake / Wake & sign in from that
console's own profile (shared useConsoleWake), and a not-answering notice
naming Settings > System > Power Saving > Features Available in Rest Mode.
- Connection Check: a closed :9021 on a console DDP says is asleep reads
"Your PS5 is in rest mode. Wake it, then check again", no VPN hints.
- PowerControl reads the shown console's credentials and seeds its power
state from the poller.
- Auto-redeploy holds while the console rests (no TCP probes); keep-awake
skips resting consoles.
- Upload queue parks an upload that meets rest mode as "Waiting for the
PS5 to wake" (pending, no spent recovery attempt) and resumes it on the
helper-up edge, desktop and browser; Files uploads stop instead of
retrying into a sleeping console and resume the same way.
- 16 new strings (joberr.console_rest_mode among them) in all 21 locales.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
When the helper's port and the loader both refuse, or the session connect
gets no TCP answer, require_in asks DDP (2 SRCH within ~800 ms). A 620
reply is the new token console_rest_mode ("The PS5 is in rest mode. Wake
it, and the helper can be sent again."). No reply after a timeout stays
ava1_unreachable with a plainer message. DDP 200 or silence after a
refusal keeps helper_not_ava1 as before. A verdict DDP explained is
cached with the pool refusal for 12 s (PROBED_REFUSAL_TTL, past the
client's 10 s poll); other refusals keep their 5 s / 30 s TTLs. The probe
is injectable (Pool::with_power_probe) and skips loopback hosts.
GET /api/ps5/status failures answer {error, reason, power}. Upload,
download and relay stalls on a console DDP calls asleep fail with
console_rest_mode instead of the bare "no durable progress".
payload_check (Tauri and the browser twin) passes reason and power
through; payloadCheck returns them as optional fields.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Engine: POST /api/ps5/helper/stop { addr } sends node.shutdown over the
paired AVA1 session (payload_lifecycle::shutdown_running_payload) and waits
for the helper's port to go quiet before answering {state: "stopped"};
{state: "not_running"} when nothing answered. Refused with 409
{reason: "busy"} while this engine runs a transfer, download, relay,
console file operation or install against that console (a per-console busy
count held by each job, plus the install slot); 502 {reason: "refused" |
"still_running"} otherwise.
Client: "Stop helper" beside Send/Resend on the Connection screen, behind a
confirm. A stop marks the console "stopped by you" (persisted in
localStorage, so desktop and browser builds keep it across restarts). The
reconnect loop, a queue's mid-run recovery and unattended queue runs (wake
resume, re-attach after restart) leave a marked console alone. Send helper,
Check, the setup wizard, Quick bring-up and starting an upload clear the
mark, as does a helper seen answering a minute after the stop. Connection
and the screen gate show "Stopped" rather than a missing helper. Strings in
all 21 locales.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Below the sm breakpoint a Files row is checkbox, icon, name (ellipsized,
full name in title), size and one ⋮ that opens the row's actions in the
desktop order with the same labels and icons, Delete last in red. The
desktop buttons and the menu come from one list (buildRowActions), so
both run the same handlers. The menu is the row's existing right-click
menu: portaled, keyboard-navigable, closed by outside tap and Escape,
placed against the ⋮ and flipped above it near the bottom of the screen.
The selection bar keeps "N selected" on one line and scrolls its actions
sideways on a phone; the Recent list sits on the opaque floating-menu
surface above the rows at every width (it was the 4% page tint, so the
rows read through it). Toolbar, drive chips, path bar, sort chips and
rows are denser at phone width; Add folder / New folder / Refresh show
their icon alone there. The desktop layout is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Engine: POST /api/folder/build unpacks a game image (or, once the pkg
reader lands, a debug-key .pkg) into <out>/<name>. It runs as a job with
Write and Verify stages; every path is checked before anything is written
(.., absolute, empty components, names this OS cannot hold, file/folder
clashes, case clashes on a case-insensitive drive); data streams in 8 MiB
chunks with cancel per chunk; the folder is written as <name>.partial/
(marked so the output-folder sweep can tell it from a user's own) and
renamed only after the read-back: every size and empty folder, and the
SHA-256 of a seeded sample (or every file) re-read from a freshly opened
source. An existing target, a plain folder source and a shortfall of
free space (files + 64 MiB) are refused before the job starts.
Client: a third "What to make" tile, A game folder. A folder source
disables Make with "This is already a folder"; .pkg is accepted as a
source; then is keep only; the finished card shows the folder and Show in
folder. The queue, prefs, pipeline and stage list know the folder kind.
Strings in all 21 locales.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RxCGb8Asp1A1AQYiv1kbgS
Pausing a routed send now ends both consoles' halves, so its journal is free
to resume from any device with any helper. The driving engine keeps a shared
record of each send on the receiving console (/data/ps5upload/c2c/<tx>.json)
with a 45 s lease refreshed every 15 s; another device lists those sends
(GET /api/transfer/ps5-to-ps5/remote), takes one over or watches a direct one
(POST /api/transfer/ps5-to-ps5 with adopt: take | watch, same tx_id), and the
engine that had it stops driving without ending anything ("taken over by").
Helper: a new caps bit, CAP_SHARED_JOBS (4). job.status / job.cancel answer
any paired device for uploads and downloads (operations and local copies stay
owner-only), and a JobOpen moves a parked upload of another device to the
opener; an attached one is still refused. SPEC §18.5-18.6.
Test connection in Send to another console runs the console's net.reach probe
from the sending console to the dial target (plus the relay legs, and an
optional real 4 KiB direct send). Tasks shows "Sends from other devices" with
Watch / Take over / Resume here.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K3cPwhp3UrzayBfyy9PDcN
native_job_notify turns true only after a successful /api/activity poll, so a
remote engine on an older version leaves job notifications with the page. The
BgStrings parity test reads the Rust file through Vite (no node types in the
client tsconfig); a clippy fix in the tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K3cPwhp3UrzayBfyy9PDcN
Files gets "Decompress to .exfat…" on a .ffpfsc again. The first build (pulled
before 7.0.23) inflated the blocks in the helper with vendored zlib, and the helper
died with SIGBUS within minutes of loading it; once the console froze. This one has
no decoder: the helper mounts the container read-only the way ShadowMount+ mounts a
.ffpfsc (LVD image type 9, read-only single-save flags, 4096-byte sectors, pfs with
mkeymode=AC), so the kernel decodes it, and copies the file inside to <name>.part,
then renames it in the same folder. The kernel lists that file rounded up to whole
64 KiB blocks; only the inode's real length is copied.
The checks from the first build stay: the container's block table is read before the
kernel sees it, a destination that exists and a drive without room for the image plus
1 GiB are refused, the size and exFAT boot sector are checked before the rename, and
Stop removes the .part and the mount. One unpack runs at a time. The mount is not in
the mount table the helper sees, so it is recorded on the /mnt tmpfs (owner pid, LVD
unit, .part) and a helper that starts after one died mid-copy releases it; a record of
a live instance is left alone.
Measured on the FW 13.60 Phat: a 315 MB image (127 MB container) in 3.5 s and a 4.5 GB
one (3.0 GB container) in 65 s, both byte-identical to the source; Stop, a takeover
mid-copy and a stale record all left no mount, LVD unit or .part behind.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K3cPwhp3UrzayBfyy9PDcN
