Preserve the immutable 6.7.0 tag ancestry and synchronize version metadata after verified npm publication. Keep newer main performance changes and the full 6.6.0 changelog.
Add opt-in string_tables_only and CLI --string-tables-only to keep selected
strings once in typed table facts. Preserve existing defaults and legacy replies;
charge aggregate output capacity for the copies actually requested.
Validate shared CLI/MCP behavior with independent XML oracles, three real APKs,
94 source tests, nine compiled boundary tests, 36 completed MCP tasks and three
CLI parity tasks. Preserve full task estimates, negative timings, resource
admission refusal and verification limitations in the measured report.
Fixes#4982
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
JSON.parse, JSONC parse, and native Claude Code keep the last duplicate
object member, and parseClientConfiguration therefore validates the last
one. The shared setup/uninstall serializer instead edited text through
jsonc-parser's modify and findNodeAtLocation, which select the first
matching member, so setup and uninstall on configurations with repeated
server tables or rea members wrote into a shadowed member and failed
readback.
Locate every JSON text edit by the effective (last) member in one place,
replacing the first-member modify path. Removing an effective member
that shadows an earlier duplicate would activate that duplicate, whatever
its ownership, so the serializer now returns a tagged shadowed_member
failure and setup and uninstall refuse before writing a backup or the
configuration.
Fixes#4952
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Claude Code shared the generic `json` client format, which the shared
configuration reader admits through JSONC parsing with comments and
trailing commas enabled. Native Claude Code reads `.claude.json` with a
strict JSON parser (after a UTF-8 BOM), so setup reported `ready`, wrote
a registration into a file the client rejects as corrupt, and doctor
reported it `aligned`.
Give Claude Code its own `claude_code` configuration format and parse it
at the owning boundary with strict JSON.parse after removing a leading
BOM. Setup, doctor, and uninstall all consume that parser, so comment or
trailing-comma documents are now refused as malformed before any backup
or write, while plain and BOM-prefixed JSON still configure. Clients that
natively accept JSONC keep the `json` format and its behavior.
Fixes#4940
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Give retained processes and runtime roots separate cleanup leases. Bind
composed failures to their exact dependencies so process retirement removes
only process labels and root deletion resolves the remaining markers.
Retain local evidence while descriptor cleanup is unresolved.
Preserve completed decoded facts and the original cancellation reason when
final cleanup itself fails, then expose that primary after successful retry.
Remove the superseded root-error composition branch after descriptor gating.
Validation: 68 focused tests across nine files, including real retained file
handles, late process handshakes, failed final cleanup, mixed child outcomes
and cleanup retry. Static, docs (five checks) and discovery (1577 files) pass.
The producer qualification scope is controlled Node/file fixtures with the
real macOS inspector and a simulated availability gate, not native pwntools
or GDB. Task tokens/CPU/process-family memory are unmeasured; no speed claim.
Fixes#4945
Keep opaque reasons at admission, queued work, command/file projection and
successful final cleanup. Normalize Node's ABORT_ERR envelope only at the
snapshot write API that produced it, using the verified original cause.
Remove downstream signal/name/identity guesses in decoder and debugger
catches, including reply-close and path-probe handling.
The initial regressions failed 15 cases with seven controls passing.
The complete final series passes 68 focused tests, static checks, docs and
discovery. Real owned Node/file fixtures use the actual macOS inspector;
only the Linux x64 availability gate is simulated. No native pwntools or
GDB qualification or performance improvement is claimed.
Refs #4945
Interpret source-bound ELF64 NT_FILE descriptors once before debugger
launch. Bound counts and pathname coverage by the descriptor, retain exact
path bytes, and refuse relative names or dot segments rather than letting
GDB resolve them against the host. Pin GDB's pathname semantics to Unix.
Preserve the decoded partial observation on refusal, basic-only recovery,
and pre-admission cancellation. Share an exact mapping fixture and replace
the old fixture boolean modes with explicit report producers.
Validation: final series has 33 passing Vitest cases plus five Python
producer controls; 38 Linux x64 provider cases are explicitly skipped on
macOS. Typecheck/lint, docs checks and test discovery pass. Real GDB16.3
and strace reproduce the relative-path escape using only an owned fixture;
compiled admission admits a real absolute-path core and refuses that
relative-path core. Full native pwndbg / Linux x64 remain unverified here.
Task token, CPU and peak-family-memory costs are unmeasured; no speed claim.
Fixes#4942
Occupy GDB's executable slot with its owned /proc/self/exe while loading a
core, then discard it before mapping queries. Select the validated decoder
architecture explicitly and restrict build-ID lookup to the owned empty path.
Preserve cancellation reasons at debugger admission.
Keep the original fixture executable present during syscall verification and
add Debian 13 / GDB 16 coverage alongside Ubuntu 24.04 in the CI definition.
Validation: controlled producer regressions and request-profile controls;
real GDB 16.3 / strace on Linux arm64 shows the old bridge opening the
recorded executable and the fix returning the same mapping without that
access. A cross-architecture gdb-multiarch probe confirms profile selection.
The plugin was controlled; full native pwndbg and Linux x64 remain unverified.
No performance improvement claimed; task tokens/CPU/family memory unmeasured.
Fixes#4914
Adapt #4732 using canonical provider identity and explicit producer recovery. Keep the valid session retained for caller-selected disposition and preserve default capability error behavior. Refs #4653.
Co-authored-by: Luigi Pinna <luigipinna3@gmail.com>
Codex and Grok client TOML was parsed with smol-toml's default reader,
which rejects any integer beyond JavaScript's safe Number range. Valid
TOML 1.0 Int64 values such as `max_bytes = 9223372036854775807` made
setup, doctor and uninstall treat the whole file as malformed.
The client TOML owner now parses with `integersAsBigInt: "asNeeded"`,
keeping safe integers as numbers and larger ones exact as bigint, and
rejects values outside signed Int64 before any write. All TOML parses in
the document owner (document read, inline string decoding and the
preserving-edit check) use that reader. The inline-table writer emits
bigint values, and setup and doctor accept a positive bigint
`startup_timeout_sec`/`tool_timeout_sec` instead of replacing or
rejecting it.
Fixes#4917
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Adapt #4044 at the shared normal/interrupted frame owner. Preserve the decoded prefix before byte-budget accounting and cover MCP delivery across framing and UTF-8 boundaries. Refs #4016.
Co-authored-by: Xander-AJ <133286098+Xander-AJ@users.noreply.github.com>
Adapt the forwarding proposal in #4652 with a required signal before optional resource diagnostics. Migrate all callers and cover actual command registration during final close. Refs #4939.
Co-authored-by: Tapraj-Pathak <165466587+Tapraj-Pathak@users.noreply.github.com>
Normalize only filter-owned selections before streamed JSON and JSONL output, and count the same normalized payload. Keep source values and strict domain JSON validation intact. Refs #4938.
Share schema keyword roles and preserve reference bases for explicit local pointer targets and uncertain intermediate scopes. Retain literal values and canonical validation. Refs #4936.
Correct stale plist and prerequisite fixtures, and replace the TOML serializer's repeated terminal-LF regex scans with a backward scan.
Keep capacity, timeout, source-preservation, backup, and idempotence coverage intact.
Fixes#4931
Preserve typed failed attempts through root admission, hashing, inventory,
nested JavaScript readers and extraction rollback. Remove the intermediate
cause/resource records that discarded canonical cleanup capabilities.
Resolve original provider errors after confirmed release while preserving
primary failures, partial observations and invalidated-handle uncertainty.
Fixes#4928
Return the canonical owner and lease from failed releases. Compose initial
cleanup failures from explicit operation outcomes and failed attempts, and
carry their authority through the shared artifact failure contract.
Keep independent unknown cleanup unresolved. Distinguish operation failures
from completed work whose close rejected. Let the scope own current child
resource labels without replacing its canonical retry lease.
Cover mixed owners, partial release, cancellation provenance, typed close
rejections and independent unknown cleanup through the producing boundaries.
Use the shared artifact file owner for kernel report reads, preserving the
physical outcome and exact retry lease through failed close and recovery.
Remove the handwritten finally path that discarded this authority.
Cover successful and failed reads with and without a caller scope, denied
retry, blocked replacement admission, and recovery of original error markers.
Adapt the separate descriptor repair proposed in #4863.
Fixes#4926
Co-authored-by: abdooo2235 <abdalrhmantheiv15@gmail.com>
Resolve unchanged Map operations through shared binding, capture, mutation,
and ambient escape facts. Reuse that proof in application and captured
bundle analysis; unresolved get/delete receivers remain inferred candidates.
Preserve repeated native operations without manufacturing receiver escapes.
Retain uncertainty for real escapes, opaque eval, and changed constructor
adder hooks. Converge stored aliases at the producing boundary.
Adapt the JavaScript changes from #4863, excluding its Linux resource fix.
Add captured-bundle parity, repeated-operation and constructor-hook coverage.
Fixes#4853
Co-authored-by: abdooo2235 <abdalrhmantheiv15@gmail.com>
Resolve lookup identity and selected/default origin together in the plist
source owner. Physical-file and conversion failures consume that fact
rather than reconstructing intent from omission of the path argument.
Selected plist artifacts resolve independently of readiness; unresolved
scope admission does not acquire an invented file-selection context.
Remove the superseded resolver and path/requested representation.
Cover direct-target CLI/MCP refusals, physical file conditions, permission
denial, default controls, unresolved scope context and cleanup retry for
both explicit path and selected-target workflows.
Fixes#4923. Related: #4849.
Validation: public CLI/MCP baseline3 failed8 controls passed; final55
passed across7files, check:fast, docs5, discovery1566 and source review.
Actual macOS26.3 plutil CLI/MCP; no Linux/Windows native qualification.
Read failed conversion diagnostics from both output streams, retaining
raw bounded excerpts, observed byte totals, exit status and truncation.
Remove only repeated path prefixes from display text and report a useful
reason when the command prints no diagnostic. Reuse native failure
capture bounds rather than expanding accepted nonzero output into errors.
Adapted from PR #4851. Fixes#4849.
Validation: six controlled boundary failures on baseline, including the
stdout-only empty-reason defect; final46 passed across6files, real macOS
CLI/MCP controls, static checks, docs5 and discovery1565. On macOS26.3
plutil uses stderr; stdout-only coverage is controlled output. Separate
selected-target classification defect is reproduced and tracked as#4923.
Co-authored-by: apitoagents <339136651+apitoagents@users.noreply.github.com>
Make XML numeric uncertainty explicit at its producing boundary, separate
from decoded IEEE NaN and signed infinities. Shared projection preserves
known values using canonical numeric labels; unresolved XML numeric
values remain explicit unknowns. Remove the superseded known-value
wrapper and spelling alias, keeping numeric interpretation with its owner.
Cover public CLI/MCP binary archives, lookalike dictionaries, authored
32/64-bit IEEE payloads, signed zero, and XML parsing/overflow uncertainty.
Document canonical labels and unrepresented NaN payload/sign details.
Fixes#4919. Related: #4906, #4852.
Validation: baseline 2 failures/5 controls through public CLI/MCP; stable
final 439 passed across25files, check:fast, docs:check and discovery.
Native macOS plutil and existing codesign controls executed; no
Linux/Windows run or real Interface Builder engine qualification.
Consume shared value and exact-number projections before JSON validation
so decoded plist scalars cannot leak as ordinary dictionaries. Carry
projection limitations into signature evidence with entitlement context.
Related: #4906. Controlled command-runner regression passed. Real
codesign refused the authored non-finite entitlement; this does not claim
real entitlement support. Existing real signature CLI controls passed.
Recognize complete CFPropertyList literals at the XML owner and retain
canonical NaN and signed infinity identity as typed scalars. Keep scalar
wrappers distinct from lookalike dictionaries, share limitation text and
document alias normalization. Incomplete markup retains its existing
compatibility fallback and limitation. Bound token matching before case
normalization using the accepted grammar.
Adapted from PR #4908. Fixes#4906.
Validation: public CLI/MCP regression baseline 2 failed, 4 passed.
Final focused validation 430 passed across 25 files, including controlled
entitlement response coverage in the following commit. Real macOS plutil
CLI, check:fast, docs:check and discovery passed. Binary value identity
remains tracked in #4919; malformed literal policy remains #4852.
Co-authored-by: Mert Kaya <47145541+mertkayacs@users.noreply.github.com>
Request epoch timestamps from capinfos and normalize them once at the
metadata boundary, preserving nanosecond precision and explicit unknowns.
Bound numeric parsing to the producer representation and cover negative
fractions, calendar limits, malformed output and MCP projection.
Adapted from PR #4909. Fixes#4907.
Validation: 47 passed, 24 platform-specific skips across five focused
files; check:fast and test discovery passed. No live Wireshark or
Linux/Windows capture execution was performed on this macOS host.
Co-authored-by: Mert Kaya <47145541+mertkayacs@users.noreply.github.com>
Preserve original RegularFileCleanupFailure capabilities when registration
inspection or maintenance planning has no caller-supplied resource scope.
Keep scoped diagnostics with the scope that already retained the descriptor.
Cover configuration and policy sources, completed and malformed-UTF8 reads,
all four exported reader/planner boundaries, exact descriptor retry after
pathname replacement, no subsequent execution, and scoped cleanup retry.
Validation: 79 tests across four focused files; the historical baseline fails
all 16 new ownership-transfer controls with eight existing/scoped controls
passing. check:fast, format:check, test discovery and six isolated completed
public doctor comparisons passed. Scoped source review found no blocker.
Real temporary descriptors and injected close rejection on macOS arm64 /
Node24.18.0; no native Windows/Linux close faults or provider verification.
Fixes#4910
Add direct controls for literal AppData roots, denied MSIX target rechecks,
and unrecognized Claude package deployments. Verify setup refuses mutation
and registration diagnostics remain invalid for unknown package readers.
Six new controls fail at 6d5281407 before the existing 9abdad335 fix; one
existing-behavior guard passes. All seven pass on current main. Keep native
Windows/app behavior distinct from macOS filesystem and platform-selected
adapter checks.
Validation: 82 tests across four configuration/CLI files, 42 header/skill/
lifecycle tests, 144 Ghidra/doctor consumer tests, check:fast, format:check,
and test-discovery verification passed. No provider/target execution.
Refs #4378, #4432, #4496, #4501, #4384, #4676
Use entry metadata to discover registration evidence independently of target
readability. Preserve failed metadata inspection as a named invalid diagnostic
instead of treating all lookup failures as absence. Remove the duplicate
boolean existence wrapper; configuration parsing remains the read owner.
Cover native permission-denied targets and metadata, dangling configuration
entries, maintenance exclusion, and public doctor output without mutation.
Validation: 73 focused tests across four files, check:fast, format:check,
and six isolated completed public doctor cost comparisons passed on macOS
arm64 / Node24.18.0. Native Windows/Linux permissions were not verified;
POSIX permission controls skip unsupported hosts and root.
Fixes#4905
Replace configuration-specific write-file-atomic delegation with a verified
input snapshot, exclusive private staging, and an explicit canonical-parent
destination. Existing files publish by rename; first-run files use an exclusive
link so a newly created destination is retained. Both consumers read back the
publication owner's selected path.
Retain descriptor and staging cleanup capabilities through recovery and host
retry. Preserve mode and ownership, remove temporary permission handoffs, and
use positional restoration writes. Keep UTF-8 diagnostics with failed cleanup.
Validation: 284 focused tests passed, one skipped across 14 files; check:fast,
format:check, docs:check (including five consistency tests), two source reviews,
and 12 completed isolated public CLI setup/uninstall cost controls passed.
Native Windows/Linux, real editor concurrency, and providers were not verified.
Final-check/publication races and same-inode concurrent writers remain unexcluded.
Fixes#4904
Replace pathname writes with the shared configuration publication owner.
Reject replacement readback and restore through the exact staged descriptor.
Retain initial-read, readback and publication cleanup owners for retry. Share
nested cleanup handling with setup without granting authority to opaque
cancellation reasons. Separate host admission from nested file cleanup gates.
Preserve uninstall file modes, including readable read-only configurations.
Hold the exact staging permission owner through writable-descriptor handoff
and failed mode restoration. Remove the superseded writeText effect contract.
Validation: 206 focused tests across 9 files, check:fast, formatting and
docs:check passed. Three uninstall replacement controls fail on baseline
7e4eed9dd while thirteen setup controls pass. Twelve completed CLI uninstall
tasks preserve equivalent outputs, config, backups and modes0400/0640.
Two read-only source reviews completed. Native Windows/Linux and real editor
races remain unverified; in-place writer exclusion is still not provided.
Closes#4902
Retain the staged file descriptor through atomic publication, readback and
recovery. Refuse detected replacements or edits and restore through that
owned descriptor instead of reopening a pathname. First-run readback failure
now retains the file for manual recovery because pathname deletion cannot
prove exact ownership. Keep both descriptor owners when nested closes fail;
opaque close rejections do not transfer ownership.
Validation: 192 focused checks in 9 files, check:fast, formatting and docs:check
passed. Three replacement regressions fail on the baseline while six existing
recovery controls pass. Six completed setup tasks preserve equivalent config
and backups. Two read-only source reviews completed on macOS arm64 / Node24.
Native Windows/Linux and real concurrent editor behavior remain unverified.
State checks do not exclude an in-place writer after the last check.
Closes#4900
Use one registration classifier for doctor, maintenance and uninstall.
Preserve unknown identity when launcher resolution fails, and compare native
paths without collapsing symlink traversal. Keep Node argv script selection
separate; ambiguous direct-launch interpretations cannot authorize removal.
Remove the superseded boolean ownership parser.
Validation: 183 focused tests passed, 2 Windows cases skipped; check:fast
and docs:check passed. Five public regressions fail on the baseline, including
foreign registration removal in an isolated fixture. Source review and authored
Node shebang controls completed on macOS with Node 24.18.0. Native Windows,
Linux and real mise upgrades remain unverified.
Closes#4895Closes#4896
Reuse supervised raw-byte capture with a 4 MiB aggregate budget. Drain package-manager output through physical close, reject oversized complete replies, and let verbose installs finish with explicitly bounded failure diagnostics. Preserve existing command resolution, selected environment/cwd, runtime prerequisites and signal policy.
Own human progress backpressure and final write acknowledgements without ending borrowed stderr. Continue draining after destination failures and observe outstanding write rejections through their error event.
Validation: 152 checks across 12 focused files, followed by a 44-check rerun including two additional no-deadline waiter controls; typecheck/lint, docs and discovery pass. The public oversized-metadata regression fails on 4f7b5fb60. Fixtures execute owned Node/fake npm/Bun commands; no actual package install or registry request. Native Windows/Linux and exhausting-output/OOM scenarios were not run.
Small completed CLI checks return identical replies and token estimates. Six sequential baseline/candidate tasks separate first-process and two warm repeats; no speedup, true-cold or large-install memory claim. Existing descendant cleanup and signal behavior is outside this output-bound fix.
Closes#4888.
Allow the existing supervised output wait to omit its deadline while preserving bounded waits and abort-listener cleanup for current consumers. Updater package-manager commands can await physical stream completion without an invented install timeout.
Refs #4888.
Normalize the unambiguous root --update alias before output handling and
through direct createCli().serve calls. Reuse the existing output-value scan
and canonical update command, including selected environment, installation
ownership, Windows-aware launch, verification, maintenance and failure facts.
Retain option data and help/check precedence. Reject remaining Incur install
requests: update:false does not disable its explicit installer in0.6.1.
Remove the superseded output-only normalizer name and update its consumers.
Validation:123focused checks/7files pass, including compiled CLI with an owned
rejecting fake npm and direct serve across5formats. Original argument owner
fails6new controls/7pass; isolated baseline fails8workflow checks/2controls
pass. Static/docs/discovery1561 pass. Both scoped source reviews are clear.
Six completed equivalent current-version workflows compare baseline canonical
update with the alias; complete replies identical. Native Windows upgrade,
real package-manager installation and full install costs remain unverified.
The first baseline escaped a Vitest-only builtin mock and installed REA in the
pinned Node global prefix. Removed the newly added package and verified original
Node/npm/Corepack tools. Re-ran using the real builtin spy with synchronized
ESM exports, an independently rejecting PATH shim and owned npm prefix/cache.
Keep heavy tests/profiling stopped. Independent updater output bounds remain
tracked in #4888 rather than changing collection policy in this routing fix.
Fixes#4883
Resolve the current named POSIX account through OS identity rather than the
caller-selected HOME/USERPROFILE. Preserve the home prefix for repeated
separators using the same joining semantics as the existing ~/ branch.
Require an absolute HERMES_HOME for unresolved names and Windows named forms
before exposing config or skill write destinations. Reuse the shared Hermes
path owner and existing diagnostics across setup, doctor, skills and uninstall.
Validation: 113 checks across seven files pass, including the compiled CLI
setup/doctor/skill/repeat/uninstall workflow over an owned profile beneath the
actual account home and a separate caller home. The isolated unchanged baseline
fails the five new checks and passes44existing checks; final repeated-separator
controls also fail that baseline. Static/docs/discovery1559 pass. Both source
reviewers clear after correcting repeated-separator composition.
Twelve completed setup dry runs compare normal and named overrides, with exact
normal-home replies unchanged. Measurements have source hashes and separate
first/warm runs; no speedup, full Hermes agent/model or native Windows/Linux
qualification claim. No provider software was installed; heavy tests/profiling
remain stopped.
Fixes#4841
Index unshadowed direct-eval sites once at the semantic owner. Distinguish
lexical rebinding, non-strict eval-introduced names, and possible mutation
of reachable objects. Reuse existing effect timing and capture proofs;
retain original definitions and module origins without inventing assignments
or parsing or executing eval payloads. Query object uncertainty conservatively
across the owning activation so aliases remain covered when names are shadowed
or leave block scope. Retain no binding-by-eval table.
Use the shared facts for values, read provenance, reference effects and child
argument shapes. Preserve indirect/optional/shadowed eval controls, unrelated
activations, immutable shapes, strict-mode bindings and prior primitive
captures. Remove obsolete reference plumbing and the source-wide eval blocker.
Validation: 306 focused checks across 16 files pass, including compiled CLI
and stdio MCP return projections, constructor uncertainty and child-process
consumers. Against the preceding SQLite commit, 9 of 23 selected checks fail
and 14 controls pass. Static/docs/test discovery (1559 files) pass. Two source
reviews found no remaining scoped blocker after alias and sloppy-name fixes.
No analyzed JavaScript, SQLite engine, heavy test or profiler was executed.
Conservative object reachability can intentionally lose exactness; large-case
opaque-query CPU and memory remain unmeasured.
Fixes#4875
Use Database spelling only to select a candidate. Reuse shared expression
provenance and indexed mutation facts to omit proven local constructors and
retain known SQLite API origins separately from uncertain names. Match ESM
default/namespace and CommonJS export forms without collapsing their identity.
Keep type-only import origins as ambiguous runtime values, including aliases.
Carry candidate interpretation through the worker schema and graph nodes and
edges with low confidence, explicit limitations and distinct identities.
Remove the name-only SQLite storage classifier. Other storage APIs retain
existing handling; PR4863's separate Map endpoint work is not integrated.
Validation: initial9negative controls fail unchangedf5fcf5216 while4positive
controls pass; final67focused checks/8files plus60provenance-consumer checks/
4files pass. Static/docs/discovery1557 pass. Two source reviewers clear after
correcting namespace constructor identity. Public CLI checks worker/schema
projection and separate exact/candidate identities with the same filename.
Twelve small completed CLI workflows preserve exact known-API replies and
remove false local-class storage facts. No speedup/large-case memory or real
SQLite execution claim. Direct eval's existing generic invalidation defect
remains tracked separately in #4875. Heavy tests/profiling remain stopped.
Additional consumer validation caught and corrected runtime import-equals
classification: require-mode import origins remain runtime values. Three new
SQLite controls cover runtime and erased import-equals forms; compiled
child-process CLI/MCP stdio workflows pass after the correction.
Fixes#4854
Keep one typed accumulation of observed decode facts instead of constructing
partial Evidence only after string projection. Record decoded client/target
before the post-command cancellation checkpoint, parsed metadata before
manifest reads, raw manifest before interpretation, and inventory prefix
before rethrowing its exact original cause. Omit unexamined fields.
Return inventory success or exact failure plus completed prefix through
Result. Keep resource-cap stops distinct from physical failures. Attach
truncation and inventory limitations at their producing stages. Validate
complete results with the named domain schema; centralize partial Evidence
assembly in the existing failure/cleanup boundary.
Validation: six authored stage controls fail unchanged34a88d6c1; 131 focused
checks across7files, check:fast, docs:check and discovery1556 pass. Covers
manifest bytes/XML, interrupted inventory, late cleanup cancellation,
denied removal/retry and existing table interruption. Two source reviewers
clear after moving the command handoff checkpoint. No separate runtime
qualification of that precise handoff race.
Public CLI manifest failure retains metadata/raw manifest with unknown
fields omitted and original failure framing unchanged. Twelve small normal
CLI tasks retain exact replies. No speedup or memory-saving claim; failure
reply estimate189 to610cl100k tokens preserves completed Evidence. No real
Apktool, large-case memory or whole-agent qualification. Heavy tests remain
stopped.
Fixes#4882
Return completed string projections with the exact failure cause in Result.
Force partial coverage and disclose the interrupted selection and remaining
unattempted selections without inventing table status or counts.
Assemble execution once from success or partial projection. Preserve the
original failure through existing ownership/cleanup classification, then
retain completed Evidence before snapshot fallback. Existing producer
observations retain precedence. Remove the throw-only projection contract.
Validation: four authored filesystem controls fail unchanged f1edf0da0;
99 focused checks across six files, check:fast and docs:check pass. Covers
late cancellation, denied workspace cleanup and retry, opaque cancellation
between tables, exact physical cause, and CLI/MCP contracts. Public CLI
failure retains completed facts; original failure framing is unchanged.
Two source reviews found no blocker.
Twelve small normal-path CLI tasks preserve exact replies. Measurements
show no speedup; failure reply grows to retain Evidence (cl100k estimate
189 to 696). No real decoder, whole-agent, or large-case memory claim.
Heavy tests and profiling remain stopped.
Fixes#4880
Reuse the named domain locales field instead of reconstructing an optional
array from its element type. Preserve min/max selection bounds through the
CLI parser and public JSON Schema while retaining the repeat-flag guidance.
The application service already validates the input; provider authority is
unchanged.
Add a public schema regression and compare advertised Ajv2020 acceptance
and CLI parser acceptance with the owner for omission, empty, valid,
maximum, oversized and invalid-element selections.
Validation: regression fails unchanged d37c34647; 21 focused checks across
4 files, check:fast and docs:check pass. Source review found no blocker.
No real-provider or performance qualification claim. Heavy tests remain
stopped.
Fixes#4876
Separate physical string-table facts from per-selection output allocation.
Use one owner for legacy and batched projections, preserving caller labels
and normalized-directory reuse. Retain complete counts and at most one
bounded table prefix for later aliases. A skipped cached alias keeps its
known counts; unexamined tables keep explicit unknown counts.
Remove the provider's duplicated parsing and status branches. Charge the
primary legacy copy before selecting output entries and derive limitation
text from the actual emitted count. Reuse the named request input type and
clarify nullable count descriptions without changing their schema types.
Have the workspace report retained raw-prefix bytes, including a stripped
UTF-8 BOM, instead of reconstructing source cost from decoded text.
Validation: 145 focused checks across 8 files; baseline count control fails
unchanged fd6da7092; check:fast, docs:check and discovery 1556 pass. Twelve
small public CLI decode workflows preserve exact normal-path replies and
request/reply token estimates. Timing variation is noise-level; no speedup,
large-case memory or real Apktool qualification claim. Heavy tests/profiling
remain stopped.
Fixes#4874
Recover completed ordinary command results when pinned Incur formatting
exceeds the runtime string limit before stdout begins. Emit a small typed
memory constraint with an explicit unknown formatted size, the runtime limit
and JSON/filter remediation. Preserve full-output framing and leave normal
CTA, human and token-window formatting with Incur. Never rerun analysis.
Capture canonical failure identity at its logging owner, including thrown
AnalysisError projections. Recovery reads compact captured facts without
revalidating or retaining large diagnostics. Keep destination failure ownership,
prevent a second stdout document and retain quiet SIGPIPE status during recovery.
Validation: 111 distinct focused tests across 10 files; check:fast,
docs:check and discovery of all 1555 test files. Sixteen authored formatter
controls fail unchanged e103ae0ce and pass with this fix. A separate closed
reader control caught and now covers recovery EPIPE handling. The heavy JSON
string-limit regression remains excluded; no large Electron reproduction ran.
Twelve small public CLI artifact inspections preserve identical raw replies
and token estimates in TOON and JSON. Timings are noise-level, with no speedup,
cold-cache, large-result memory or real-provider qualification claim. Sentinel
and generator protocols keep their existing ownership.
Fixes#4806
morluto