Commit Graph
5170 Commits
Author SHA1 Message Date
Ysopwnableandgensokyo-dev 4ea24b6f8e refactor(cli): derive build catalogs from the REA command registry
Read CLI catalogs and registrar ownership from the REA command records, type-check the catalog helper and consolidate registrar types. Preserve the generated catalogs, mapping and CLI parity.

Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-11-16 18:57:12 +08:00
Ysopwnableandgensokyo-dev 7d502ddb0d refactor(cli)!: own builtin refusals and retire per-command MCP
Own builtin refusal paths and Markdown schemas in REA. Reject retired per-command MCP before loading command registrars, preserve invoked identity, and apply token controls to the complete selected failure document.

BREAKING CHANGE: rea <command> --mcp is removed. Start REA's MCP server with rea mcp or rea --mcp.

Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-11-16 18:26:02 +08:00
Ysopwnableandgensokyo-dev 230db70fb4 feat(gdb): bind sessions to an optional executable
Hash and load an optional initial executable before GDB readiness, without running it. Preserve initial identity and completed hash diagnostics through cancellation and cleanup retry; share the file owner with Rizin.

Fixes #11581. Refs #11682.

Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-11-16 16:19:33 +08:00
a10037ef10 refactor(cli): report unknown commands and the update notice from REA (#11692)
* refactor(cli): render help pages with REA's help owner

REA now renders the root page, every command page, help after an unknown command, the mcp group page, the mcp doctor builtin page and the completions page before incur sees argv, with incur's precedence (update check, --update without help, --mcp and the LLM manifests first; then the integrations; then --version; then root and command pages). help.ts gains arguments, custom global options, the root-only flags and integrations, and a command-list page.

The --format row on the root and command pages now lists csv and tsv (#10250), the approved difference; the mcp and mcp add pages keep the document formats they accept. Every other help byte is unchanged: 397 pages compared with incur's Help for all 196 registered commands.

Fixes #10250

* test(cli): elide the root command list wherever the root page is printed

The golden renderer chose root-page elision by case name, so help after an unknown command froze the whole command list. It now elides any stdout that opens with the root page's rea@<version> title.

* refactor(cli): answer --version, --schema and the LLM manifests in REA

src/cli/discovery.ts answers help, --version, --schema, --llms and --llms-full with incur's precedence, reading the arguments as the dispatcher receives them (after result output has adapted them, so token counts and buffering are unchanged). commandManifests.ts renders the Markdown index and full manifest, the structured manifests and the --schema document; commandSuggestion.ts ranks Did-you-mean candidates. Both are adapted from incur under its MIT licence.

--schema --format md stays with incur: incur renders it without Markdown escaping, unlike REA's renderer, and that difference awaits a decision on #11654. Compared with incur's serve for every registered command and format, 3746 of 3945 discovery invocations are identical; the rest are the 196 Markdown schemas left to incur and the three root help pages with the approved csv/tsv row.

* test(cli): pin which discovery document wins when several are asked for

Goldens for --llms with --help, --llms with --version, --llms with --llms-full, --schema with --version and --schema with --help, recorded with incur's rendering on the base branch.

* refactor(cli): report unknown commands and the update notice from REA

src/cli/updateNotice.ts owns the terminal update notice: the same trigger (terminal stdout, no explicit format), text, cache file ($XDG_CACHE_HOME/incur/updates/<package>.json), daily detached refresh through rea --incur-update-check, and opt-outs (NO_UPDATE_NOTIFIER, npm_config_update_notifier=false, CI). Discovery answers the detached refresh first, and reports an unknown command with incur's Did-you-mean suggestion, corrected invocation, help command and notice, as text on a terminal or as the error document, full-output envelope, token count or token window otherwise. outputEnvelope.ts renders these dispatcher failures.

Unknown commands with --format md stay with incur pending the Markdown escaping decision on #11654. Compared with incur's serve, 224 of 224 unknown-command invocations are identical, piped and on a terminal with a due notice.

* fix(cli): redact unknown-command guidance before rendering

Share the producing reply boundary for root lookup, unknown schemas and MCP subcommands. Preserve ordinary argument identity and existing MCP suggestion policies; omit sensitive or unsafe replay commands and redact full-output command metadata.

Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>

---------

Co-authored-by: gensokyo-dev <gensokyo-dev@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-11-16 14:46:48 +08:00
Ysopwnableandgensokyo-dev ab16163061 test(hopper): audit producer-bound snapshots through MCP
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-11-16 06:40:11 +01:00
morluto a8cd584ccf fix(cli): render help in REA and list csv and tsv formats (#11688)
* refactor(cli): render help pages with REA's help owner

REA now renders the root page, every command page, help after an unknown command, the mcp group page, the mcp doctor builtin page and the completions page before incur sees argv, with incur's precedence (update check, --update without help, --mcp and the LLM manifests first; then the integrations; then --version; then root and command pages). help.ts gains arguments, custom global options, the root-only flags and integrations, and a command-list page.

The --format row on the root and command pages now lists csv and tsv (#10250), the approved difference; the mcp and mcp add pages keep the document formats they accept. Every other help byte is unchanged: 397 pages compared with incur's Help for all 196 registered commands.

Fixes #10250

* test(cli): elide the root command list wherever the root page is printed

The golden renderer chose root-page elision by case name, so help after an unknown command froze the whole command list. It now elides any stdout that opens with the root page's rea@<version> title.

* refactor(cli): answer --version, --schema and the LLM manifests in REA (#11690)

* refactor(cli): answer --version, --schema and the LLM manifests in REA

src/cli/discovery.ts answers help, --version, --schema, --llms and --llms-full with incur's precedence, reading the arguments as the dispatcher receives them (after result output has adapted them, so token counts and buffering are unchanged). commandManifests.ts renders the Markdown index and full manifest, the structured manifests and the --schema document; commandSuggestion.ts ranks Did-you-mean candidates. Both are adapted from incur under its MIT licence.

--schema --format md stays with incur: incur renders it without Markdown escaping, unlike REA's renderer, and that difference awaits a decision on #11654. Compared with incur's serve for every registered command and format, 3746 of 3945 discovery invocations are identical; the rest are the 196 Markdown schemas left to incur and the three root help pages with the approved csv/tsv row.

* test(cli): pin which discovery document wins when several are asked for

Goldens for --llms with --help, --llms with --version, --llms with --llms-full, --schema with --version and --schema with --help, recorded with incur's rendering on the base branch.
2026-11-16 12:55:30 +08:00
morluto 6c4e2bba9d test(cli): register decompile-all once in the overnight CLI schema registry (#11689)
registerCoreBinaryCommands already registers decompile-all; the test registered it a second time. incur overwrote the duplicate silently; REA's command table (#11680) rejects it, so every case in the file failed on main.
2026-11-16 12:34:19 +08:00
morluto 5257a3d912 refactor(cli): read global arguments once with one REA owner (#11687)
parseCliGlobalArguments replaces normalizeCliArguments, splitIncurBuiltinArguments and splitCliTimingArguments: it normalizes spaced output forms, routes the root --update alias, splits builtin switches and output values, and reads the timing global in one pass. The output-argument reader, the launcher's doctor path, lazy registration, the Amp scope guard, completions, mcp add, strict positionals and sensitive-value redaction read that record instead of scanning argv again. Only the incur hand-off adapter remains until dispatch moves to REA. Output is unchanged; the CLI parity goldens pass as recorded.
2026-11-16 12:23:51 +08:00
morluto ec1cf3a9bd refactor(cli): render command help with REA's own help owner (#11683)
* refactor(cli): render command help with REA's own help owner

Add src/cli/help.ts and move the three incur Help.formatCommand callers
(rea mcp add, rea mcp doctor, rea mcp --http) onto it. Option spelling and
the unwrapped field type move to src/cli/optionFields.ts, which completion
now reads instead of its own copies. A counted incur-import guard records
the names still imported from incur so the list can only shrink.

Refs #11654

* refactor(cli): share option spelling across CLI owners and count every incur import

The argv parser, strict positionals, contract command projection and the
command registry now read the flag spelling, declared field type and count
marker from src/cli/optionFields.ts instead of their own copies. The
incur-import guard also counts namespace, default, star, side-effect and
dynamic imports, so `import * as Incur from "incur"` cannot bring a removed
owner back.

* refactor(cli): format and filter CLI output without incur (#11684)

* refactor(cli): format and filter CLI output without incur

formatCliOutput encodes TOON (@toon-format/toon), JSON, YAML and JSON Lines
itself instead of calling incur's Formatter; YAML still loads on first use.
src/cli/outputFilter.ts owns the --filter-output grammar and returns a
Result: an unclosed slice is a parse failure, so output validation reads
the parser instead of a second bracket scanner, and the parsed paths
travel with the output arguments to the result writer and JSON Lines.
@toon-format/toon moves from devDependencies to dependencies because REA
now imports it at runtime.

Refs #11654

* test(cli): pin output encodings and filter grammar with literal cases

Literal expected outputs for empty values, TOON scalars, objects and rows,
JSON-text re-indentation, nested BigInt values, JSON Lines rows and YAML,
and literal parsed paths and unclosed-slice refusals for --filter-output,
so a change to either owner fails without comparing against itself.
2026-11-16 12:01:04 +08:00
morluto f14e9a1d28 refactor(cli): register every command in an REA-owned command table (#11680)
* refactor(cli): register every command in an REA-owned command table

* fix(cli): flag grouped fields and named selectors in each command's input binding
2026-11-16 11:39:07 +08:00
morluto dbdbda3d07 refactor(cli): bind command argv with REA's own parser (#11681)
* refactor(cli): bind command argv with REA's own parser

* test(cli): assert complete option field errors from the argv parser
2026-11-16 11:37:52 +08:00
morluto 7af53a1b42 refactor(desktop): pass the desktop capture's observers, session runner and store through a port (#11676)
The desktop application capture workflow constructed the desktop provider's
X11 window observer and content store and called its session runner,
application inventory, program lookup and home-folder walker directly.
src/ports/DesktopApplicationSession.ts now names them as
DesktopCaptureProvider, with the plan, outcome, request and walk types they
exchange; src/desktop/DesktopCaptureProvider.ts implements it and the CLI
passes it in the capture host. application -> desktop entry imports: 20 -> 14.

Refs #11583
2026-11-15 22:28:05 -05:00
morluto 80c60ac68c test(cli): cover terminal stdout behaviour in the CLI parity goldens (#11677)
* test(cli): cover terminal stdout behaviour in the CLI parity goldens

* test(cli): run the terminal command as the owned process and relay its output live
2026-11-16 11:26:56 +08:00
morluto 81fe44d7d8 refactor(evidence): build a completed provider operation's Evidence in one owner (#11673)
* test(adapters): compare CLI and MCP Evidence for every CLI analysis operation

Run each official, workflow, native, artifact and managed operation that
has a CLI command through the CLI command tree and the MCP server over
one simulated provider, and require identical Evidence or failure. The
differences main has today are recorded per operation so the W2 invoke
path can only remove them.

Split bindDirectAnalysis(dependencies) out of createDirectAnalysis so the
CLI can run over injected session owners.

* test(adapters): give each parity case its own CLI and MCP connection

* refactor(evidence): build a completed provider operation's Evidence in one owner

The CLI's DirectAnalysis and the MCP provider-operation handler each
defaulted the subject, admitted the completed analysis and created the
Evidence. completedOperationEvidence in SessionEvidenceReuse now does it
for both, reusing the record the session issued for the same execution
and parameters. The session-Evidence lookup is no longer exported.

* test(evidence): deliver the record a one-shot Ghidra decompile session issued
2026-11-15 22:25:52 -05:00
0f6b330cb6 refactor(ghidra): classify Ghidra file failures through the shared provider classifier (#10873)
* fix(pcap): report a refused tool launch as an unavailable tool

When a capture read's tool launch was refused with EPERM, pcap reported it
as unreadable tool output (AnalysisProtocolError) instead of an unavailable
tool. Pcap also never told "not executable" apart from "missing".

Add classifyLaunchFailure in src/process as the one owner of which spawn
errnos mean missing (ENOENT, ENOTDIR) and which mean present but not
executable (EACCES, EPERM, ENOEXEC, EISDIR). Pcap, apktool and gdb used three
different errno lists; all three now use it.

* refactor(process): classify owned command outcomes through one owner

Add ProviderOutcome, read once from the supervisor's facts, and classifyProviderOutcome, the one mapping from an outcome to a typed error. Migrate pcap and the shared artifact producer command failure (TShark, sigrok). Add the rea/provider-failure-classifier lint rule with a shrink-only baseline of per-site (cause: unknown) => AnalysisError functions.

* refactor(process): classify Java and adb launch refusals with the shared owner (#10598)

The Ghidra and JADX Java probes and adb's version probe now take
missing-versus-not-executable from classifyLaunchFailure instead of their
own errno checks. ELOOP and ENAMETOOLONG join the owner's missing set, so
adb keeps its existing classification and both Java probes now treat a
non-directory path component as a missing runtime.

* fix(pcap): preserve command launch attribution

* test(pcap): expect an unknown failing stage when a limited capture tool exits unsuccessfully

* refactor(process): classify owned command outcomes through one owner

Add ProviderOutcome, read once from the supervisor's facts, and classifyProviderOutcome, the one mapping from an outcome to a typed error. Migrate pcap and the shared artifact producer command failure (TShark, sigrok). Add the rea/provider-failure-classifier lint rule with a shrink-only baseline of per-site (cause: unknown) => AnalysisError functions.

* test(lint): cover baselined files in the provider-failure classifier rule

* fix(apktool): report the decoder JVM's heap exhaustion as memory, not a target refusal

The supervisor now watches for a managed runtime's out-of-memory report while output streams (JVM java.lang.OutOfMemoryError, V8 'JavaScript heap out of memory'), so a retention budget cannot hide it. ProviderOutcome gains a runtime-exhausted kind and classifyProviderOutcome maps it once. Apktool decode opts in and moves to the shared classifier; Ghidra's single-purpose output token becomes the JVM runtime option and its area matching reuses the shared table.

* fix(process): keep the runtime's out-of-memory report on the memory error it supports

* fix(pcap): attribute a limited capture's exit through the prlimit wrapper

* fix(process): keep a shortened out-of-memory line marked as shortened

* fix(process): identify the refused executable for every launcher shape

Name the executable on errnos Node throws synchronously from spawn
(ENOEXEC, ENOTDIR, ...) and read the Windows native launcher's
executable-launch constraint, so pcap attributes those refusals to the
selected tool or limiter instead of an unknown stage.

* test(pcap): prove synchronous launch naming with ENOTDIR on every POSIX host

* fix(apktool): advise the JVM limit that governs the reported memory area

* fix(ghidra): read the supervisor's parsed OOM area instead of reparsing lines

* fix(sqlite,capture): read V8 heap exhaustion from streamed output, not retained stderr (#10724)

The SQLite worker and the Node capture decoders now declare runtime: "v8", so the supervisor observes V8's heap report as output streams. Both classifiers read the runtime-exhausted outcome instead of matching retained stderr, so a report beyond the retention budget, or one whose process then overflowed its output budget, is still reported as memory exhaustion.

* refactor(evm): classify EVMole worker outcomes through the shared provider classifier (#10716)

* refactor(evm): classify EVMole worker outcomes through the shared provider classifier

The EVMole worker's cancellation, deadline, output budget and CPU/file-size signals now go through classifyProviderOutcome; only the prlimit launch refusal and the reserved EFBIG exit remain in its dialect. The worker is opted into V8 exhaustion scanning so a heap report names its fixed 256 MiB heap. providerOutcomeOf now reads a launch refusal only from a spawn errno, so a private-storage or carrier read errno is no longer reported as a limiter launch failure.

* fix(evm): keep a stream failure's exit status and update Linux worker expectations

* refactor(process): read launch refusals through the launcher identity owner

* refactor(wasm): classify WABT command outcomes through the shared provider classifier (#10727)

* refactor(wasm): classify WABT command outcomes through the shared provider classifier

WABT's cancellation, command deadline, output budget, and inherited-limit signals now go through classifyProviderOutcome; its dialect keeps only the configuration-probe and launch-refusal unavailability and the unclassified-exit output error. A tool ended by SIGXCPU/SIGXFSZ or the memory limit is now a resource constraint instead of an output error, and a refused launch in the producer phase is reported as unavailable. The shared timeout context gains the analysis setting that selected the deadline, and the output-limit context may name its exceeded budget.

* fix(wasm): keep setup guidance when a WABT version probe's output cannot be collected

* refactor(providers): classify raw provider failures through one owner

* refactor(process): build classified cancellations with cancelledBy and the carried caller reason

* chore(lint): count runOwned's close-failure mapper from main in the classifier baseline

* fix(process): carry the runtime's exhaustion report on every memory error that claims it

SQLite and historical-capture memory errors recognised a streamed V8 heap report but dropped it, so a report outside retained output reached callers without its source line or omission flags. One projection, runtimeExhaustionReport, now owns that shape for the shared classifier and both consumers. Also lowers the EVMole and WABT cancellation-constructor baselines.

* fix(process): keep the provider's recovery for a native-thread exhaustion report

The shared classifier replaced the provider-supplied recovery with generic capacity advice whenever the JVM reported it could not create a native thread, dropping Apktool's launcher guidance and flipping retryable to true. The provider's recovery now wins; the capacity advice remains the fallback for providers that supply none.

* refactor(pwntools): classify pwntools failures through the shared provider classifier (#10782)

* refactor(pwntools): classify pwntools failures through the shared provider classifier

pwntools' per-site classifier is replaced by pwntoolsFailureDialect for classifyProviderFailure. Command outcomes go through classifyProviderOutcome, launch refusals through launcherRefusalOf instead of a 'spawn ' syscall prefix check, the input limit through the file dialect's fileSize, and the cleanup report through commandCleanup. The shared read cancellation now carries the reader's partial observation, and resource-signal reports accept provider detail.

* fix(pwntools): report a failed command's output once and expect resource_constraint for overflow

The shared outcome owner already attaches the command's output as capturedOutput, so the pwntools adapter fallback no longer copies the same text into diagnostics.stdout, diagnostics.stderr and diagnostics.captured_output; diagnostics keep the exit facts. Update the overflow expectations the dialect migration missed (layout provider cleanup and over-budget rows, pwntools cleanup) to resource_constraint.

* refactor(wasm): leave WABT cancellation to the shared owners

WabtFailures matched any Error named AbortError as a cancellation. No WABT path produces one: reads, path checks and owned commands report cancellation as ArtifactReaderFailure or OwnedCommandFailure, which classifyProviderFailure and classifyProviderOutcome already classify with the caller's reason. The provider's two signal checks now build the cancellation with cancelledBy. Lowers the AnalysisCancelledError constructor baseline for both WABT files to zero.

* fix(wasm): report a cancelled snapshot write as the caller's cancellation

The private snapshot write passed the caller's signal to Node's writeFile, whose abort envelope reached the dialect's unclassified output failure once the name-based AbortError guess was removed. The write now turns the envelope carrying its own signal's reason into cancelledBy at the producer.

* refactor(firmware): classify firmware failures through the shared provider classifier (#10810)

Firmware producers now report their own deadline expiry: the start check, the output inventory and launcher admission throw the phase timeout where they observe the abort, and a read the phase signal cancelled reports it through the dialect's interruption. The classifier no longer rereads a cancellation's reason as a timeout. A new readCleanup hook keeps firmware's cleanup report shape.

* refactor(captures): classify TShark, sigrok and pcap failures through the shared provider classifier

TsharkCaptureProvider and SigrokRecordingProvider replace #failure with a dialect, and PcapProvider replaces analysisFailure and the tool-identity classifier. ArtifactAnalysisResources exposes its command context and a readCleanupFailure composer so the readCleanup hook keeps each provider's cleanup report shape.

* refactor(jeb): classify JEB failures through the shared provider classifier

JEB's raw MCP failures go through classifyProviderFailure with a JEB dialect;
a typed attribution step reports a shared phase's failure under the REA
operation. Each MCP phase deadline now reports its own interruption when it
aborts, deciding a cancellation's owner from the caller's own signal instead
of comparing abort reasons. A refusal carrying the JVM's OutOfMemoryError
report becomes a resource constraint naming JEB's Java heap.

* fix(jeb): advise the JVM limit that governs the reported memory area

* refactor(ghidra): classify Ghidra failures through the shared provider classifier

PSP extension inspection reports through classifyProviderFailure with a
Ghidra dialect. The startup deadline decides at abort time whether the shared
startup owner closed it, so Ghidra and Hopper no longer compare the abort
reason with the owner's sentinel, and Ghidra's startup no longer compares a
producer cancellation's reason with the deadline's. Three helpers that only
construct one error kind now name that kind instead of AnalysisError.

* refactor(hopper): classify original-source read failures through the shared provider classifier (#10838)

* test(ghidra): pin the aborted startup deadline over a producer cancellation

* refactor(kuna): classify Kuna failures through the shared provider classifier

Session operations, the workspace copy and the project export report raw
failures through classifyProviderFailure with Kuna dialects; the copy and
export keep their cleanup report shapes through the readCleanup hook.

* refactor(adb): report a pulled file's digest failure from its own type

* refactor(sqlite): classify worker-reply read failures through the shared provider classifier

* refactor(ida): classify IDA operation failures through the shared provider classifier

* refactor(process): let artifact-resource consumers classify through the shared provider classifier (#10867)

* test(providers): pin attribution of a cleanup-failed read's typed primary

* refactor(sqlite): classify database failures through the shared provider classifier

* refactor(providers): keep a cancelled read's partial observation and provider diagnostics

* refactor(ghidra): classify Ghidra file failures through the shared provider classifier

* refactor(ghidra): classify Ghidra failures through the shared provider classifier (#10836)

* refactor(ghidra): classify Ghidra failures through the shared provider classifier

PSP extension inspection reports through classifyProviderFailure with a
Ghidra dialect. The startup deadline decides at abort time whether the shared
startup owner closed it, so Ghidra and Hopper no longer compare the abort
reason with the owner's sentinel, and Ghidra's startup no longer compares a
producer cancellation's reason with the deadline's. Three helpers that only
construct one error kind now name that kind instead of AnalysisError.

* refactor(hopper): classify original-source read failures through the shared provider classifier (#10838)

* test(ghidra): pin the aborted startup deadline over a producer cancellation

* refactor(kuna): classify Kuna failures through the shared provider classifier (#10843)

Session operations, the workspace copy and the project export report raw
failures through classifyProviderFailure with Kuna dialects; the copy and
export keep their cleanup report shapes through the readCleanup hook.

* chore(guards): shrink the cancellation construction baseline after merging the base

* refactor(kuna): report a retained installation read's cleanup from its reader type (#10877)

* refactor(kuna): report a retained installation read's cleanup from its reader type

* refactor(process): share the JVM option that governs each reported memory area (#10881)

* refactor(ida): classify IDA operation failures through the shared provider classifier (#10859)

* refactor(ida): classify IDA operation failures through the shared provider classifier

* refactor(process): let artifact-resource consumers classify through the shared provider classifier (#10867)

* test(providers): pin attribution of a cleanup-failed read's typed primary

* fix(javascript): read analysis workers' V8 heap exhaustion from streamed output, not retained stderr (#10890)

* refactor(kuna): report a retained installation read's cleanup from its reader type

* refactor(process): share the JVM option that governs each reported memory area (#10881)

* fix(javascript): read analysis workers' V8 heap exhaustion from streamed output, not retained stderr

* test(browser): pin web-bundle heap exhaustion reported past the retained stderr budget

* refactor(process): recognise a signal's own abort through one owner (#10896)

* refactor(process): recognise a signal's own abort through one owner

* chore(verify): lower the direct cancellation construction baselines the stack's migrations removed

* refactor(sqlite): read the owned worker's heap ceiling from one limit (#10907)

* refactor(sqlite): classify the owned worker's outcomes through the shared provider classifier (#10912)

* refactor(process): call the signal abort owner directly and match NaN reasons

* chore(verify): lower the SQLite cancellation construction baseline the worker migration removed

* feat(rizin): report a shell refusal as a typed refusal record beside rejected completion (#10925)

* refactor(kuna): report installation inspection failures through one Kuna dialect (#10913)

* refactor(errors): let Hopper errors carry their own presentation so the domain presenter no longer imports them (#10944)

* refactor(process): recognise the remaining command aborts through the signal owner (#10906)

* refactor(process): recognise the remaining command aborts through the signal owner

* test(process): cover a NaN abort reason on both execFileOutput cancellation paths

* fix(process): keep a successful execFile callback when the caller aborts with a null reason

* test(process): abort execFileOutput while a released descendant still holds its pipes

* test(process): join the late-abort command in finally so a failed readiness check still releases its descendant

* refactor(errors): move Hopper errors to src/hopper and drop their tags from the domain union (#10964)

* refactor(errors): move Hopper errors to src/hopper and drop their tags from the domain union

* refactor(process): recognise SQLite, WABT and Flutter aborts through the signal owner (#10975)

* refactor(kuna): report doctor installation failures through the shared Kuna installation dialect (#10978)

* refactor(process): recognise SQLite, WABT and Flutter aborts through the signal owner

* refactor(kuna): report doctor installation failures through the shared Kuna installation dialect

* fix(kuna): keep installation facts on a composed read-and-close failure

* style(hopper): format the moved Hopper lease and cache publisher fixture

* fix(sqlite): count only the caller's own abort as a SQLite inspection cancellation (#11018)

* fix(ida): count only the caller's own abort of the busy-health pause as cancellation (#11013)

* test(kuna): assert the cleanup-incomplete doctor check by code and facts, not its wording

* refactor(errors): report an unrecognised command outcome with the outcome the classifier examined (#11003)

* refactor(errors): report an unrecognised command outcome with the outcome the classifier examined

* fix(errors): add a dialect's unclassified diagnostics to the context's instead of replacing them

* fix(evm): count only the caller's own abort as an EVMole inspection cancellation (#11011)

* fix(server): report only startup's own abort as the caller's HTTP listen cancellation (#11017)

* refactor(errors): classify domain errors from one table and generate the error catalogue from it (#11060)

docs(errors): generate the error catalogue from the classification tables

* refactor(errors): report unrecognised pcap, WABT and producer-command outcomes with the examined outcome (#11044)

* refactor(errors): report unrecognised pcap, WABT and producer-command outcomes with the examined outcome

* fix(pcap): keep the packet summary when only the DNS wire pass reports an unrecognised outcome

* refactor(ida): require the host platform for IDA MCP connections and snapshot the child environment with it (#11073)

* refactor(verify): count plain throws in parser-folder process-lifecycle owners (#11071)

JadxWorkspaceRoot and PwntoolsDecoder are exempt from parser-plain-throw because their plain throws report a private helper's process failures. The exemption left later throws in those files unchecked. A counted lifecycle-plain-throw guard now records their current throws (4 and 1); the baseline may only shrink, so a parse fault or defect added there must use ParseFailureError or invariantViolation.

* fix(native): report a signature inspection defect as an adapter failure (#11069)

withSignatureTarget wrapped every throw from the inspection callback as
AnalysisOutputError("Native output parsing failed"), so an
InternalInvariantError (a REA defect) was reported as malformed native
command output. A defect now leaves the owned work, so runOwned releases
the private signature copy first and reports the defect as the native
provider's adapter failure. If releasing the copy also fails, the defect
stays primary and the cleanup failure joins it through the shared cleanup
composition (analysisErrorWithCleanupFailure + bindCleanupDependencies).

* refactor(verify): give the javascript verifiers' roots to VerifierFixture (#11068)

* refactor(verify): give the javascript verifiers' roots to VerifierFixture

VerifierFixture.retain(reason) keeps a root that an owned command whose
cleanup is incomplete may still use, replacing each verifier's local
retain flag and cleanup closure.

* refactor(verify): give the Ghidra COM, DOS, 6502, NativeAOT and raw verifiers' workspaces to VerifierFixture (#11070)

* refactor(verify): give the Ghidra COM, DOS, 6502, NativeAOT and raw verifiers' workspaces to VerifierFixture

* refactor(verify): keep the Ghidra decompiler-failure, decompile-timeout and no-return workspaces through VerifierFixture until owned cleanup is verified (#11077)

* feat(artifacts): surface the named integrity check in public artifact errors (#11052)

* feat(artifacts): surface the named integrity check in public artifact errors

ArtifactOperationError now carries integrityCheck from the reader failure
and projects it as details.integrity_check, including with artifact
integrity details. Rebuild sites copied a different subset of reader facts
each (ProcessCli.artifactFailureFacts, withCleanup, desktop, file-operation
cleanup, resource export, DMG inventory, safe output tree); they now spread
ArtifactReaderFailure.primaryFacts(), so a new reader fact has one owner.
ArtifactIntegrityCheck moves to the domain beside ArtifactAdmissionLimit.

* test(artifacts): check rebuilt cleanup failures keep the named integrity check

* refactor(adb): report a pulled file's digest failure from its own type (#10848)

* fix(pcap): report a refused tool launch as an unavailable tool

When a capture read's tool launch was refused with EPERM, pcap reported it
as unreadable tool output (AnalysisProtocolError) instead of an unavailable
tool. Pcap also never told "not executable" apart from "missing".

Add classifyLaunchFailure in src/process as the one owner of which spawn
errnos mean missing (ENOENT, ENOTDIR) and which mean present but not
executable (EACCES, EPERM, ENOEXEC, EISDIR). Pcap, apktool and gdb used three
different errno lists; all three now use it.

* refactor(process): classify owned command outcomes through one owner

Add ProviderOutcome, read once from the supervisor's facts, and classifyProviderOutcome, the one mapping from an outcome to a typed error. Migrate pcap and the shared artifact producer command failure (TShark, sigrok). Add the rea/provider-failure-classifier lint rule with a shrink-only baseline of per-site (cause: unknown) => AnalysisError functions.

* refactor(process): classify Java and adb launch refusals with the shared owner (#10598)

The Ghidra and JADX Java probes and adb's version probe now take
missing-versus-not-executable from classifyLaunchFailure instead of their
own errno checks. ELOOP and ENAMETOOLONG join the owner's missing set, so
adb keeps its existing classification and both Java probes now treat a
non-directory path component as a missing runtime.

* fix(pcap): preserve command launch attribution

* test(pcap): expect an unknown failing stage when a limited capture tool exits unsuccessfully

* refactor(process): classify owned command outcomes through one owner

Add ProviderOutcome, read once from the supervisor's facts, and classifyProviderOutcome, the one mapping from an outcome to a typed error. Migrate pcap and the shared artifact producer command failure (TShark, sigrok). Add the rea/provider-failure-classifier lint rule with a shrink-only baseline of per-site (cause: unknown) => AnalysisError functions.

* test(lint): cover baselined files in the provider-failure classifier rule

* fix(apktool): report the decoder JVM's heap exhaustion as memory, not a target refusal

The supervisor now watches for a managed runtime's out-of-memory report while output streams (JVM java.lang.OutOfMemoryError, V8 'JavaScript heap out of memory'), so a retention budget cannot hide it. ProviderOutcome gains a runtime-exhausted kind and classifyProviderOutcome maps it once. Apktool decode opts in and moves to the shared classifier; Ghidra's single-purpose output token becomes the JVM runtime option and its area matching reuses the shared table.

* fix(process): keep the runtime's out-of-memory report on the memory error it supports

* fix(pcap): attribute a limited capture's exit through the prlimit wrapper

* fix(process): keep a shortened out-of-memory line marked as shortened

* fix(process): identify the refused executable for every launcher shape

Name the executable on errnos Node throws synchronously from spawn
(ENOEXEC, ENOTDIR, ...) and read the Windows native launcher's
executable-launch constraint, so pcap attributes those refusals to the
selected tool or limiter instead of an unknown stage.

* test(pcap): prove synchronous launch naming with ENOTDIR on every POSIX host

* fix(apktool): advise the JVM limit that governs the reported memory area

* fix(ghidra): read the supervisor's parsed OOM area instead of reparsing lines

* fix(sqlite,capture): read V8 heap exhaustion from streamed output, not retained stderr (#10724)

The SQLite worker and the Node capture decoders now declare runtime: "v8", so the supervisor observes V8's heap report as output streams. Both classifiers read the runtime-exhausted outcome instead of matching retained stderr, so a report beyond the retention budget, or one whose process then overflowed its output budget, is still reported as memory exhaustion.

* refactor(evm): classify EVMole worker outcomes through the shared provider classifier (#10716)

* refactor(evm): classify EVMole worker outcomes through the shared provider classifier

The EVMole worker's cancellation, deadline, output budget and CPU/file-size signals now go through classifyProviderOutcome; only the prlimit launch refusal and the reserved EFBIG exit remain in its dialect. The worker is opted into V8 exhaustion scanning so a heap report names its fixed 256 MiB heap. providerOutcomeOf now reads a launch refusal only from a spawn errno, so a private-storage or carrier read errno is no longer reported as a limiter launch failure.

* fix(evm): keep a stream failure's exit status and update Linux worker expectations

* refactor(process): read launch refusals through the launcher identity owner

* refactor(wasm): classify WABT command outcomes through the shared provider classifier (#10727)

* refactor(wasm): classify WABT command outcomes through the shared provider classifier

WABT's cancellation, command deadline, output budget, and inherited-limit signals now go through classifyProviderOutcome; its dialect keeps only the configuration-probe and launch-refusal unavailability and the unclassified-exit output error. A tool ended by SIGXCPU/SIGXFSZ or the memory limit is now a resource constraint instead of an output error, and a refused launch in the producer phase is reported as unavailable. The shared timeout context gains the analysis setting that selected the deadline, and the output-limit context may name its exceeded budget.

* fix(wasm): keep setup guidance when a WABT version probe's output cannot be collected

* refactor(providers): classify raw provider failures through one owner

* refactor(process): build classified cancellations with cancelledBy and the carried caller reason

* chore(lint): count runOwned's close-failure mapper from main in the classifier baseline

* fix(process): carry the runtime's exhaustion report on every memory error that claims it

SQLite and historical-capture memory errors recognised a streamed V8 heap report but dropped it, so a report outside retained output reached callers without its source line or omission flags. One projection, runtimeExhaustionReport, now owns that shape for the shared classifier and both consumers. Also lowers the EVMole and WABT cancellation-constructor baselines.

* fix(process): keep the provider's recovery for a native-thread exhaustion report

The shared classifier replaced the provider-supplied recovery with generic capacity advice whenever the JVM reported it could not create a native thread, dropping Apktool's launcher guidance and flipping retryable to true. The provider's recovery now wins; the capacity advice remains the fallback for providers that supply none.

* refactor(pwntools): classify pwntools failures through the shared provider classifier (#10782)

* refactor(pwntools): classify pwntools failures through the shared provider classifier

pwntools' per-site classifier is replaced by pwntoolsFailureDialect for classifyProviderFailure. Command outcomes go through classifyProviderOutcome, launch refusals through launcherRefusalOf instead of a 'spawn ' syscall prefix check, the input limit through the file dialect's fileSize, and the cleanup report through commandCleanup. The shared read cancellation now carries the reader's partial observation, and resource-signal reports accept provider detail.

* fix(pwntools): report a failed command's output once and expect resource_constraint for overflow

The shared outcome owner already attaches the command's output as capturedOutput, so the pwntools adapter fallback no longer copies the same text into diagnostics.stdout, diagnostics.stderr and diagnostics.captured_output; diagnostics keep the exit facts. Update the overflow expectations the dialect migration missed (layout provider cleanup and over-budget rows, pwntools cleanup) to resource_constraint.

* refactor(wasm): leave WABT cancellation to the shared owners

WabtFailures matched any Error named AbortError as a cancellation. No WABT path produces one: reads, path checks and owned commands report cancellation as ArtifactReaderFailure or OwnedCommandFailure, which classifyProviderFailure and classifyProviderOutcome already classify with the caller's reason. The provider's two signal checks now build the cancellation with cancelledBy. Lowers the AnalysisCancelledError constructor baseline for both WABT files to zero.

* fix(wasm): report a cancelled snapshot write as the caller's cancellation

The private snapshot write passed the caller's signal to Node's writeFile, whose abort envelope reached the dialect's unclassified output failure once the name-based AbortError guess was removed. The write now turns the envelope carrying its own signal's reason into cancelledBy at the producer.

* refactor(firmware): classify firmware failures through the shared provider classifier (#10810)

Firmware producers now report their own deadline expiry: the start check, the output inventory and launcher admission throw the phase timeout where they observe the abort, and a read the phase signal cancelled reports it through the dialect's interruption. The classifier no longer rereads a cancellation's reason as a timeout. A new readCleanup hook keeps firmware's cleanup report shape.

* refactor(captures): classify TShark, sigrok and pcap failures through the shared provider classifier

TsharkCaptureProvider and SigrokRecordingProvider replace #failure with a dialect, and PcapProvider replaces analysisFailure and the tool-identity classifier. ArtifactAnalysisResources exposes its command context and a readCleanupFailure composer so the readCleanup hook keeps each provider's cleanup report shape.

* refactor(jeb): classify JEB failures through the shared provider classifier

JEB's raw MCP failures go through classifyProviderFailure with a JEB dialect;
a typed attribution step reports a shared phase's failure under the REA
operation. Each MCP phase deadline now reports its own interruption when it
aborts, deciding a cancellation's owner from the caller's own signal instead
of comparing abort reasons. A refusal carrying the JVM's OutOfMemoryError
report becomes a resource constraint naming JEB's Java heap.

* fix(jeb): advise the JVM limit that governs the reported memory area

* refactor(ghidra): classify Ghidra failures through the shared provider classifier

PSP extension inspection reports through classifyProviderFailure with a
Ghidra dialect. The startup deadline decides at abort time whether the shared
startup owner closed it, so Ghidra and Hopper no longer compare the abort
reason with the owner's sentinel, and Ghidra's startup no longer compares a
producer cancellation's reason with the deadline's. Three helpers that only
construct one error kind now name that kind instead of AnalysisError.

* refactor(hopper): classify original-source read failures through the shared provider classifier (#10838)

* test(ghidra): pin the aborted startup deadline over a producer cancellation

* refactor(kuna): classify Kuna failures through the shared provider classifier

Session operations, the workspace copy and the project export report raw
failures through classifyProviderFailure with Kuna dialects; the copy and
export keep their cleanup report shapes through the readCleanup hook.

* refactor(adb): report a pulled file's digest failure from its own type

* refactor(ghidra): classify Ghidra failures through the shared provider classifier (#10836)

* refactor(ghidra): classify Ghidra failures through the shared provider classifier

PSP extension inspection reports through classifyProviderFailure with a
Ghidra dialect. The startup deadline decides at abort time whether the shared
startup owner closed it, so Ghidra and Hopper no longer compare the abort
reason with the owner's sentinel, and Ghidra's startup no longer compares a
producer cancellation's reason with the deadline's. Three helpers that only
construct one error kind now name that kind instead of AnalysisError.

* refactor(hopper): classify original-source read failures through the shared provider classifier (#10838)

* test(ghidra): pin the aborted startup deadline over a producer cancellation

* refactor(kuna): classify Kuna failures through the shared provider classifier (#10843)

Session operations, the workspace copy and the project export report raw
failures through classifyProviderFailure with Kuna dialects; the copy and
export keep their cleanup report shapes through the readCleanup hook.

* chore(guards): shrink the cancellation construction baseline after merging the base

* refactor(ida): let IDA close failures read the session's later close (#10811)

* refactor(errors): read cleanup completion from dependencies instead of pushing it to dependents

AnalysisError.cleanupIncomplete becomes a getter over the error's own unresolved state and its bound dependencies. The WeakRef dependents graph and resolveCleanup's push propagation are removed, and subclasses mark their initial state through markCleanupIncomplete instead of overriding a public mutable field.

* fix(errors): settle bound dependencies before binding more

A composite whose dependencies had released but whose status was never read
stayed incomplete when a later dependency was bound.

* refactor(errors): let errors read a lease group's release instead of resolving them (#10759)

AnalysisCleanupLeaseGroup no longer calls resolveCleanup() on its error.
The error holds the group as a release claim and reads cleanupReleased when
asked, so completion follows the lease owner without a push.

* chore(verify): count resolveCleanup and bindCleanupDependencies calls so they only shrink (#10763)

* refactor(errors): let errors read a lease group's release instead of resolving them

AnalysisCleanupLeaseGroup no longer calls resolveCleanup() on its error.
The error holds the group as a release claim and reads cleanupReleased when
asked, so completion follows the lease owner without a push.

* chore(verify): count resolveCleanup and bindCleanupDependencies calls so they only shrink

* refactor(process): let worker cleanup errors read their root epoch's release

OwnedWorkerRuntime's root epoch is now an AnalysisCleanupReleaseClaim. Errors bound to it read its release when asked instead of the epoch pushing resolveCleanup() through weak references, and a runtime without a retained root binds an already-released owner. Removes the OwnedWorkerCleanupFailure structural type and four resolveCleanup() pushes; the counted baseline shrinks accordingly.

* refactor(pcap): let pcap cleanup errors read command and snapshot release (#10792)

Retained pcap commands, retained snapshot failures and snapshot completion descriptors now own their release and errors read it: commands and snapshot failures implement AnalysisCleanupReleaseClaim, and a failed completion-descriptor close is retried through an AnalysisCleanupLease bound to its error. Removes four resolveCleanup() pushes; the counted baseline shrinks accordingly.

* refactor(process): let Darwin helper cleanup errors read their owner's release

Darwin helper cache and run-token reader root cleanup now record an AnalysisCleanupRelease that every failure reports; a successful retry confirms it. Replaces the first-failure anchor that later failures depended on and the resolveCleanup() push. Removes two resolveCleanup() and two bindCleanupDependencies() calls; both counted baselines shrink.

* refactor(ghidra): let Ghidra cleanup errors read the client's and checkpoint's release

GhidraClient and GhidraCheckpoint bind each cleanup failure to an AnalysisCleanupRelease and confirm it when a later close or retry releases everything, instead of keeping a set of failures and pushing resolveCleanup() to each. Removes two resolveCleanup() calls; the counted baseline shrinks.

* refactor(process): let debugger session cleanup errors read a confirmed stop

OwnedDebuggerSessionRegistry binds each retained gdb or rizin cleanup failure to the session's AnalysisCleanupRelease and confirms it when a stop is confirmed, keeping the latest failure for retry reporting, instead of pushing resolveCleanup() to a set of failures. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(process): drop the debugger registry's unused closeOperation option

Returning the latest retained failure removed the last reader of closeOperation; remove the option and the values gdb and rizin supplied.

* refactor(server): let environment shutdown errors read a successful close

serveEnvironment binds each shutdown failure to an AnalysisCleanupRelease that a later successful close confirms, instead of pushing resolveCleanup() to a set of failures. The state-file retry test now also requires the earlier report to read the release. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(desktop): release retained screen-judge cleanup through its lease

Each retained judge or screenshot cleanup owner is wrapped in an AnalysisCleanupLease bound to its failure; close() releases the lease, which records the release the failure reads, instead of closing the owner and pushing resolveCleanup(). Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(ida): let IDA close failures read the session's later close

IdaSessionClient's close failure binds an owner that is both its resource owner and its release claim, read from the client's closed state, instead of keeping weak references and pushing resolveCleanup() after a successful retry. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(javascript): let source-map codec cleanup errors read a later close (#10814)

* refactor(javascript): let source-map codec cleanup errors read a later close

OwnedCodec binds each close failure to an AnalysisCleanupRelease that a later successful close confirms, replacing the first-failure anchor that later failures depended on and the resolveCleanup() push. Removes one resolveCleanup() and one bindCleanupDependencies() call; both counted baselines shrink.

* refactor(cutter): let pending Cutter command errors read the completion receipt (#10815)

* refactor(cutter): let pending Cutter command errors read the completion receipt

Each pending Cutter command keeps an AnalysisCleanupRelease bound to its timeout failure; the producer's terminal completion receipt confirms it instead of pushing resolveCleanup(). Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(application): let provider candidate cleanup errors read the candidate's close (#10817)

* refactor(application): let provider candidate cleanup errors read the candidate's close

AnalysisProviderRegistry keeps one AnalysisCleanupRelease per candidate whose resource close is unconfirmed; failures bind to it and a later confirmed close confirms it, instead of pushing resolveCleanup() to a set of failures. Removes one resolveCleanup() call; the counted baseline shrinks.

* fix(errors): bind a cleanup release claim to an error at most once

AnalysisError kept release claims in an array, so a provider whose retried close returned the same error bound the same claim again on every retry and each cleanupIncomplete read scanned the duplicates. Keep the claims in a set.

* refactor(application): let reverse-tool cleanup errors read a verified stop (#10819)

* refactor(application): let reverse-tool cleanup errors read a verified stop

Each retained provider process group in ReverseEngineeringService keeps an AnalysisCleanupRelease; its command and retry failures bind to it and a later verified stop confirms it, instead of collecting failures and pushing resolveCleanup(). Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(sqlite): let SQLite worker and root cleanup errors read their release (#10821)

Each retained SQLite snapshot cleanup records an AnalysisCleanupRelease for its worker and for its runtime root; the resource failures bind to them and a verified stop or root removal confirms them, instead of pushing resolveCleanup(). Composite failures keep reading those resource failures. Removes two resolveCleanup() calls; the counted baseline shrinks.

* refactor(application): let JADX and session close failures read their owner's release (#10854)

* refactor(android): let JADX cleanup errors read their retained cleanup's release

Each retained JADX cleanup records an AnalysisCleanupRelease for the admission failure's descriptor owners and one for the whole retained cleanup. The admission failure and every error the cleanup reports bind to them, and a confirmed close confirms them, instead of pushing resolveCleanup() through WeakRefs to reported errors. Removes two resolveCleanup() calls; the counted baseline shrinks.

* refactor(application): let session close failures read the client's later close

A retained session cleanup records an AnalysisCleanupRelease; unowned cleanup failures the client's close reported bind to it and the successful close confirms it, instead of the session collecting them and pushing resolveCleanup(). Failures that already follow their own owner keep reading it. Adds a regression for an unowned close failure. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(browser): parse Runtime and Debugger events once by registered schemas (#10742)

* refactor(browser): parse frame lifecycle events once by a registered event schema

CdpConnection admits each event through protocol/CdpEvents.ts, which parses
the five Page frame lifecycle events once into identity facts plus a frame
URL observation, like a frame-tree node. Screenshot, page capture, the
observation timeline, WebMCP, Electron inspection and the execution window
read those facts instead of re-parsing the raw payload; CdpFrameFacts no
longer carries the raw source record. Playwright CDP session subscriptions
name an event from Playwright's protocol table, and an architecture guard
counts the remaining raw event payload reads so they can only shrink.

* refactor(browser): parse Runtime and Debugger events once by registered schemas

Register Debugger.scriptParsed, Debugger.scriptFailedToParse,
Runtime.executionContextCreated, Runtime.executionContextDestroyed and
Runtime.consoleAPICalled. Script and context payloads carry both the
tolerant facts passive capture uses and the active runtime's strict
declaration, derived from one parse, plus the declaration's byte size for
the runtime retention budget. Page capture, Electron script capture, the
V8 Inspector provider and the runtime source inventory now read the
registered payload. The cdp-raw-event-params baseline drops from 20 to 7.

* refactor(browser): admit transport events through the registered-event owner

Remove the unused cdpFrameUrl wrapper; its fragment cases now exercise
observeCdpFrameUrl. Add a connection-level test that two subscribers share
one parsed payload and the raw params.

* fix(browser): report an unserializable event declaration's size instead of throwing

JSON.stringify can throw on a declaration JSON.parse accepted (nesting
deeper than the serializer's stack). The registry sized script and context
declarations inside the connection's WebSocket callback, so such an event
escaped as an uncaught exception. The size is now a Result; the runtime
source inventory raises the failure inside its own catch, as before.

* refactor(browser): parse Network events once by registered schemas (#10743)

* refactor(browser): parse Network events once by registered schemas

Register Network.requestWillBeSent, responseReceived, loadingFinished,
loadingFailed, webSocketCreated and webSocketFrameSent/Received. The
request payload carries the tolerant facts page capture and the timeline
read plus the active runtime's strict declaration. Page capture's network
and WebSocket handlers, the timeline's request tracking and the runtime
request inventory now read the registered payload. The
cdp-raw-event-params baseline drops from 7 to 5.

* refactor(browser): drop payload helpers the capture handlers no longer use

* refactor(browser): read every CDP event payload through the registry (#10745)

* refactor(browser): read every CDP event payload through the registry

Register Target.attachedToTarget, Target.detachedFromTarget,
Page.lifecycleEvent and WebMCP.toolsAdded/toolsRemoved, and carry a
requested navigation's reason and a commit's transition type in the
frame-event payload. The timeline no longer threads raw params into its
entries: each entry receives its timestamp, frame and loader identity
explicitly. WebMCP discovery, the service-worker policy blocker and the
target-event matcher read the registered payload. With no raw reads left,
the cdp-raw-event-params guard is enforced at zero instead of counted.

* refactor(browser): drop raw payload helpers the timeline and worker blocker no longer use

* refactor(browser): type Playwright CDP session events by the registry (#10751)

PlaywrightCdpSession.on subscribes only to registered events and hands
the listener the payload parsed by its registered schema, returning the
unsubscribe function. Private storage seeding reads registered payloads,
including the events of non-flat child sessions, and applies its strict
seeding policy to the parsed facts. Its local attach, detach, nested
message and pause schemas and the storage-only script/context schemas are
removed; Debugger.paused, Runtime.executionContextsCleared and
Target.receivedMessageFromTarget join the registry.

* test(browser): admit fake policy events through the event registry

The service-worker policy now reads event.registered, but the PlaywrightScenarioSession fake connection delivered raw events, so three detached-target tests never reached the policy. The fake now admits each event with receiveCdpEvent as CdpConnection does.

* fix(browser): fail the CDP session closed when an event listener throws (#10758)

* fix(browser): fail the CDP session closed when an event listener throws

An exception from a CDP event subscriber escaped CdpTransportCore's socket message callback uncaught (for example the service-worker blocker's protocol_error on a Target event without a sessionId). The transport now stops delivery, fails the connection as a protocol error whose cause is the listener's exception, rejects pending and later commands with it, and notifies protocol-failure subscribers. The Playwright CDP session relay applies the same rule to its event listeners.

* fix(browser): report a failed service-worker policy connection at the next policy check

The policy's event listener rejects a malformed Target event, which now fails its connection. With no policy command pending, nothing reported it and later iframe targets went unprotected. The policy records its connection's protocol failure and assertBlocked, run before each scenario step, rethrows it.

* fix(browser): keep cleanup commands available after a listener failure

A listener failure no longer marks the wire as broken. The transport still rejects pending commands, notifies protocol-failure subscribers, stops delivering events and refuses reuse, but it keeps reading replies and admits commands sent in cleanup grace. The service-worker policy can then remove its init scripts, restore the main realm, detach and release its socket. Wire-level protocol errors still refuse every later command. The PlaywrightScenarioSession policy fake gains onProtocolFailure, which the policy now subscribes to.

* refactor(browser): keep raw CDP event payloads out of the received event (#10764)

* refactor(browser): keep raw CDP event payloads out of the received event

CdpReceivedEvent and CdpTransportEvent no longer carry params: receiveCdpEvent hands the raw payload only to the registered schema, so a subscriber cannot read it by property access, destructuring or computed keys. This supersedes the event.params text guard in architecture-guards, which is removed.

* refactor(browser): read registered event payloads in the JavaScript verifiers

The runtime, storage and URL-userinfo verifiers subscribe to CdpConnection, which no longer hands subscribers raw params. They now read event.registered. A test scans every scripts/ onEvent callback for raw payload reads, since the TypeScript check does not cover them. The storage verifier's pause trace records a malformed pause instead of the unregistered reason member.

* refactor(browser): track the timeline's main frame with the document tracker (#10779)

TimelineCapture kept its own mutable main frame id and replaced it on a root replacement. It now holds the CdpDocumentTracker that CdpTargetSession owns and observes the replacement document through it; every main-frame check (lifecycle, navigation, document request, redirect) reads the tracker's current frame. A characterization test pins that the replaced root's lifecycle events are no longer recorded.

* refactor(browser): select CDP events only by their registered name (#10829)

CdpReceivedEvent carried the wire method as a plain string, so a subscriber
could compare it with a misspelt or unregistered name and the check compiled
but never matched. Three owners selected Inspector.targetCrashed, an
unregistered event, that way.

Register Inspector.targetCrashed and drop the wire method from the received
event. Subscribers select events by registered.method, whose type is the
registered event names. The capture coverage start keeps reading the
wire domain, now the event's domain field.

* fix(browser): retire a policy child session that CDP reports gone

After a listener failure the connection stops delivering events, so an
iframe's Target.detachedFromTarget never retires its policy target. Cleanup
then sent Page.removeScriptToEvaluateOnNewDocument to the dead session,
kept the stale child, and every retry failed before detaching the root.

The service-worker policy owner now treats CDP's session-not-found
rejection as proof that the child is gone and retires it. The classifier
moves from CdpTargetSession to CdpCommandRejection, its owner, and the
target-session detach path uses the same one.

* refactor(browser): route target events through one session owner (#10835)

* refactor(browser): route target events through one session owner

Page capture, WebMCP discovery, screenshot capture and the runtime sources
each compared the envelope session ID themselves, while the observation
window and execution collection used cdpTargetEventMatches, which also
routes the root Target.detachedFromTarget that names the target's session.

onCdpTargetEvent in CdpTargetEvents.ts now subscribes every target consumer
with that one rule; the matcher is private and the per-subscriber
comparisons are removed. Electron inspection keeps its own filter while
its capture isolation awaits a maintainer decision.

* fix(browser): report WebMCP subframe tools whose document is unverified (#10839)

* fix(browser): report WebMCP subframe tools whose document is unverified

A committed subframe navigation kept the frame's tools silently when CDP
reported no loader at the same URL, and removed them on any URL change even
when the loader proved the same document.

Each frame now carries a CdpDocumentTracker, the session's document
identity owner. Only a proven change removes the frame's tools. An unknown
comparison keeps them visible, marks webmcp_tools attach-limited and names
the frames whose document identity could not be established.

* test(browser): prove an unverified live subframe limits WebMCP coverage

* fix(browser): key a WebMCP registration by its frame and tool name

A subframe tool kept across a committed navigation that changed the URL
but not the loader was stored under a key built from the old URL, so a
later toolsAdded for the same frame and name added a second entry.
Registrations are now keyed by frame and name; tool_key still records the
registering frame URL.

* test(browser): deliver fake CDP events through the transport's admission (#10840)

Event fakes each rebuilt the received event: CdpAgentHints typed its
listeners as raw wire events behind an unknown cast, and the Electron,
scenario-session and routing tests each wrapped receiveCdpEvent by hand.

fakeCdpEvents (tests/fixtures) is a CdpConnection.onEvent fake whose emit
admits each raw event with receiveCdpEvent and the wire byte count, so a
change to the received event shape fails at the fake instead of silently
not exercising subscribers.

* refactor(browser): own the selected page generation in one target document tracker (#10860)

* refactor(browser): own the selected page generation in one target document tracker

Page capture, screenshot, the observation window, the runtime session and
the execution window each built a document tracker or rule of their own
from root-frame navigation events. CdpTargetDocument now follows the target
session's root-frame events and the frame-tree replies its consumers read,
classifies root identity once, and reports each navigation to consumers
that decide only what it means for their evidence.

* fix(browser): report a lost scenario policy connection before the next step (#10864)

* fix(browser): report a lost scenario policy connection before the next step

The service-worker policy recorded only protocol failures. When its CDP
socket closed or its page session detached, later documents lost the
blocking script, but the scenario learned of it only when a frame check
found an unprotected realm. The owner now records the disconnect or
detach when CDP reports it; the next policy check and the next step's
action report it first.

* fix(browser): refuse an unregistered CDP command before it reaches the browser (#10868)

JavaScript verifiers can name any command, and the send owners refused an
unregistered one only when its reply arrived, after the browser had acted
on it. CdpConnection and the Playwright session now refuse the name before
sending, and the Playwright session refuses an unregistered event name
instead of subscribing to events no listener receives. The verifier text
scan stays as an early warning and documents what it cannot see.

* fix(browser): keep a proven root change and the loaderless selection's uncertainty

Two review findings on the target document owner. A root event naming
another frame with a malformed loader set `changed` but was announced as
unidentified, so the execution window failed with protocol_error and
dropped what it had collected; the owner now announces it as the proven
change it is. The runtime session compared each frame-tree reply with the
tracker's current document, which can learn a loader from an event, so a
loaderless selection followed by an event and a reply with that loader
compared as same; it now compares the reply with the selected document
through the owner, and reports target_changed as before.

* test(browser): refuse the next scenario step after the policy session detaches

* refactor(browser): read WebMCP document identity from the target document owner (#10884)

* refactor(browser): read WebMCP document identity from the target document owner

WebMCP discovery compared the first and last frame-tree roots itself and
kept a document tracker per subframe. CdpTargetDocument now follows the
selected root and each committed subframe for it: the final frame tree
feeds the owner, a root change proven by a navigation event during
discovery is target_changed as in page capture, and a subframe whose
committed document changed loses its tools. A subframe navigation without
a loader still keeps the tools and reports them as unverified.

* fix(browser): start a subframe first seen within its document with an unknown loader

A subframe absent from the initial frame tree and first seen through Page.navigatedWithinDocument had no document tracker, so a later commit started a fresh one and the tools it registered meanwhile lost the unverified-document limitation. The owner now starts such a subframe with an unknown loader.

* fix(browser): refuse a verifier's unregistered CDP event name at subscription (#10885)

JavaScript verifiers subscribed to every event with CdpConnection.onEvent
and compared registered.method with a string literal. A misspelt or
unregistered name never matched and the verifier waited until its timeout.
CdpConnection.onRegisteredEvent now takes one event name, refuses an
unregistered one before subscribing (the same registry check the Playwright
session applies), and hands the listener only that event's registered
parse. The verifiers subscribe only through it, and the evaluation test
rejects a raw onEvent subscription in scripts.

* fix(browser): keep subframe documents by frame and loader without their URLs

* refactor(domain): give strict UTF-8 decoding one owner with an explicit BOM policy (#10728)

* refactor(domain): give strict UTF-8 decoding one owner with an explicit BOM policy

* fix(domain): keep the leading-BOM behaviour every migrated UTF-8 decoder had

* refactor(domain): read the domain layer's strict UTF-8 through the BOM-policy owner (#10803)

Twelve domain decoders built their own fatal TextDecoder with ignoreBOM: true. They now use strictUtf8Decoder("preserve"), which keeps that behaviour: each reads bytes where a leading U+FEFF is content (binary-format names and fields, JSON and XML read byte-for-byte, recorded bodies, registry seeds whose kept bytes are re-emitted). The counted baseline drops from 74 to 62.

* refactor(artifacts): read artifact readers' strict UTF-8 through the BOM-policy owner (#10871)

Fourteen artifact reader modules (17 sites) built their own fatal TextDecoder. They now take it from strictUtf8Decoder with the policy each had: every one kept a leading BOM (ignoreBOM: true), so they use "preserve"; JavaScriptArtifactFiles keeps it for JavaScript and HTML entries and strips it otherwise, as before. The strict-utf8-decoder baseline drops those 17 sites.

* refactor(domain): drop TextDecoder imports left by the strict UTF-8 migration

* refactor(application): read MCP registration metadata through the strict UTF-8 owner

* refactor(formats): read binary-format readers' strict UTF-8 through the BOM-policy owner (#10903)

* refactor(formats): read binary-format readers' strict UTF-8 through the BOM-policy owner

* refactor(formats): drop the model readers' unused TextDecoder imports

* refactor(browser): read browser capture and fetch strict UTF-8 through the BOM-policy owner (#10935)

* refactor(browser): read browser capture and fetch strict UTF-8 through the BOM-policy owner

* refactor(native): read LLDB and Mach-O name strict UTF-8 through the BOM-policy owner (#10936)

* refactor(native): read LLDB and Mach-O name strict UTF-8 through the BOM-policy owner

* refactor(process): read MCP and NDJSON transport frames' strict UTF-8 through the BOM-policy owner (#10937)

* refactor(process): read MCP and NDJSON transport frames' strict UTF-8 through the BOM-policy owner

* refactor(application): read JSON and file inputs' strict UTF-8 through the BOM-policy owner (#10938)

* refactor(application): read JSON and file inputs' strict UTF-8 through the BOM-policy owner

* refactor(providers): read IDA, Electron, Ghidra and reference strict UTF-8 through the BOM-policy owner (#10939)

* refactor(providers): read IDA, Electron, Ghidra and reference strict UTF-8 through the BOM-policy owner

* refactor(evm): read the hex bytecode carrier's strict UTF-8 through the owner and refuse new local decoders (#10940)

* refactor(evm): read the hex bytecode carrier's strict UTF-8 through the BOM-policy owner

* chore(architecture): match local strict UTF-8 decoders in any option order

* fix(il2cpp): resolve ELF ABS64 and GLOB_DAT image definitions (#11075)

Derive bounded ELF ABS64 and GLOB_DAT link-image values for IL2CPP module
scanning while preserving the default conservative relocation policy.
Retain ambiguous hash declarations, validate complete hash-table mappings,
and disclose runtime-binding limitations in CLI and MCP evidence.

Validation: 408 focused tests, check:fast and docs:check passed. Public
CLI/MCP and independently linked ARM64/x86-64 ELF arithmetic were checked;
the original Unity binaries remain unavailable.

Refs #10478
Related: #10495

Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>

* fix(il2cpp): read ELF hash-table words through BinaryReader (#11083)

* test(pcap): check the retained DNS-pass summary by coverage, not limitation wording

* refactor(reverse): fake objdump and Rizin commands only through the owned supervisor (#11076)

* refactor(reverse): fake objdump and Rizin commands only through the owned supervisor

* test(reverse): fake launch refusals and failed exits as the owned supervisor reports them

* refactor(reverse): name objdump and Rizin command repairs through the command port (#11080)

* refactor(desktop): let build-computer and agentenv cleanup errors read their owner's release (#10855)

* refactor(errors): read cleanup completion from dependencies instead of pushing it to dependents

AnalysisError.cleanupIncomplete becomes a getter over the error's own unresolved state and its bound dependencies. The WeakRef dependents graph and resolveCleanup's push propagation are removed, and subclasses mark their initial state through markCleanupIncomplete instead of overriding a public mutable field.

* fix(errors): settle bound dependencies before binding more

A composite whose dependencies had released but whose status was never read
stayed incomplete when a later dependency was bound.

* refactor(errors): let errors read a lease group's release instead of resolving them (#10759)

AnalysisCleanupLeaseGroup no longer calls resolveCleanup() on its error.
The error holds the group as a release claim and reads cleanupReleased when
asked, so completion follows the lease owner without a push.

* chore(verify): count resolveCleanup and bindCleanupDependencies calls so they only shrink (#10763)

* refactor(errors): let errors read a lease group's release instead of resolving them

AnalysisCleanupLeaseGroup no longer calls resolveCleanup() on its error.
The error holds the group as a release claim and reads cleanupReleased when
asked, so completion follows the lease owner without a push.

* chore(verify): count resolveCleanup and bindCleanupDependencies calls so they only shrink

* refactor(process): let worker cleanup errors read their root epoch's release

OwnedWorkerRuntime's root epoch is now an AnalysisCleanupReleaseClaim. Errors bound to it read its release when asked instead of the epoch pushing resolveCleanup() through weak references, and a runtime without a retained root binds an already-released owner. Removes the OwnedWorkerCleanupFailure structural type and four resolveCleanup() pushes; the counted baseline shrinks accordingly.

* refactor(pcap): let pcap cleanup errors read command and snapshot release (#10792)

Retained pcap commands, retained snapshot failures and snapshot completion descriptors now own their release and errors read it: commands and snapshot failures implement AnalysisCleanupReleaseClaim, and a failed completion-descriptor close is retried through an AnalysisCleanupLease bound to its error. Removes four resolveCleanup() pushes; the counted baseline shrinks accordingly.

* refactor(process): let Darwin helper cleanup errors read their owner's release

Darwin helper cache and run-token reader root cleanup now record an AnalysisCleanupRelease that every failure reports; a successful retry confirms it. Replaces the first-failure anchor that later failures depended on and the resolveCleanup() push. Removes two resolveCleanup() and two bindCleanupDependencies() calls; both counted baselines shrink.

* refactor(ghidra): let Ghidra cleanup errors read the client's and checkpoint's release

GhidraClient and GhidraCheckpoint bind each cleanup failure to an AnalysisCleanupRelease and confirm it when a later close or retry releases everything, instead of keeping a set of failures and pushing resolveCleanup() to each. Removes two resolveCleanup() calls; the counted baseline shrinks.

* refactor(process): let debugger session cleanup errors read a confirmed stop

OwnedDebuggerSessionRegistry binds each retained gdb or rizin cleanup failure to the session's AnalysisCleanupRelease and confirms it when a stop is confirmed, keeping the latest failure for retry reporting, instead of pushing resolveCleanup() to a set of failures. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(process): drop the debugger registry's unused closeOperation option

Returning the latest retained failure removed the last reader of closeOperation; remove the option and the values gdb and rizin supplied.

* refactor(server): let environment shutdown errors read a successful close

serveEnvironment binds each shutdown failure to an AnalysisCleanupRelease that a later successful close confirms, instead of pushing resolveCleanup() to a set of failures. The state-file retry test now also requires the earlier report to read the release. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(desktop): release retained screen-judge cleanup through its lease

Each retained judge or screenshot cleanup owner is wrapped in an AnalysisCleanupLease bound to its failure; close() releases the lease, which records the release the failure reads, instead of closing the owner and pushing resolveCleanup(). Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(ida): let IDA close failures read the session's later close

IdaSessionClient's close failure binds an owner that is both its resource owner and its release claim, read from the client's closed state, instead of keeping weak references and pushing resolveCleanup() after a successful retry. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(javascript): let source-map codec cleanup errors read a later close (#10814)

* refactor(javascript): let source-map codec cleanup errors read a later close

OwnedCodec binds each close failure to an AnalysisCleanupRelease that a later successful close confirms, replacing the first-failure anchor that later failures depended on and the resolveCleanup() push. Removes one resolveCleanup() and one bindCleanupDependencies() call; both counted baselines shrink.

* refactor(cutter): let pending Cutter command errors read the completion receipt (#10815)

* refactor(cutter): let pending Cutter command errors read the completion receipt

Each pending Cutter command keeps an AnalysisCleanupRelease bound to its timeout failure; the producer's terminal completion receipt confirms it instead of pushing resolveCleanup(). Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(application): let provider candidate cleanup errors read the candidate's close (#10817)

* refactor(application): let provider candidate cleanup errors read the candidate's close

AnalysisProviderRegistry keeps one AnalysisCleanupRelease per candidate whose resource close is unconfirmed; failures bind to it and a later confirmed close confirms it, instead of pushing resolveCleanup() to a set of failures. Removes one resolveCleanup() call; the counted baseline shrinks.

* fix(errors): bind a cleanup release claim to an error at most once

AnalysisError kept release claims in an array, so a provider whose retried close returned the same error bound the same claim again on every retry and each cleanupIncomplete read scanned the duplicates. Keep the claims in a set.

* refactor(application): let reverse-tool cleanup errors read a verified stop (#10819)

* refactor(application): let reverse-tool cleanup errors read a verified stop

Each retained provider process group in ReverseEngineeringService keeps an AnalysisCleanupRelease; its command and retry failures bind to it and a later verified stop confirms it, instead of collecting failures and pushing resolveCleanup(). Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(sqlite): let SQLite worker and root cleanup errors read their release (#10821)

Each retained SQLite snapshot cleanup records an AnalysisCleanupRelease for its worker and for its runtime root; the resource failures bind to them and a verified stop or root removal confirms them, instead of pushing resolveCleanup(). Composite failures keep reading those resource failures. Removes two resolveCleanup() calls; the counted baseline shrinks.

* refactor(android): let JADX cleanup errors read their retained cleanup's release

Each retained JADX cleanup records an AnalysisCleanupRelease for the admission failure's descriptor owners and one for the whole retained cleanup. The admission failure and every error the cleanup reports bind to them, and a confirmed close confirms them, instead of pushing resolveCleanup() through WeakRefs to reported errors. Removes two resolveCleanup() calls; the counted baseline shrinks.

* refactor(application): let session close failures read the client's later close

A retained session cleanup records an AnalysisCleanupRelease; unowned cleanup failures the client's close reported bind to it and the successful close confirms it, instead of the session collecting them and pushing resolveCleanup(). Failures that already follow their own owner keep reading it. Adds a regression for an unowned close failure. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(desktop): let build-computer and agentenv cleanup errors read their owner's release

Build-computer teardown, retained task readers, remote reader folders and agentenv solve sandboxes each record an AnalysisCleanupRelease that their bound failures read; a verified teardown or stop confirms it instead of pushing resolveCleanup() through WeakRef sets. Long-run job failures read whether every job they name has confirmed its stop, so a failure naming two jobs no longer reads complete when only one stops (regression added). WeakRef sets kept only where late facts (completion receipts, reader views) are still retained as diagnostics. Removes six resolveCleanup() calls; the counted baseline shrinks.

* refactor(kuna): let source-map and Kuna cache cleanup errors read their owner's release (#10856)

* refactor(browser): let late source-map fetch errors read their response's release

A pending source-map fetch and a response cancellation each record an AnalysisCleanupRelease that their cancellation or deadline failure reads; releasing the late response or the reader lock confirms it instead of pushing resolveCleanup(). Removes two resolveCleanup() calls; the counted baseline shrinks.

* refactor(kuna): let project-cache cleanup errors read each owner's release

Each retained Kuna resource owner (process, staging, hash descriptor, retired file) records a release that the errors reporting it read; an error completes once every owner it reports confirms, and retirement that moves staging to a retained file forwards the staging owner's release to the file's. Replaces the WeakRef report list, its pruning and the resolveCleanup() push. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(process): forbid resolveCleanup() pushes; native markers read their supervisor's release (#10857)

* refactor(providers): let an attributed client failure read its source's completed cleanup

Re-attributing a failure whose source cleanup already completed now binds a claim reading that source's release instead of pushing resolveCleanup() onto the new error; a pending source is still bound as a dependency. Resources stay as history, as before. Removes one resolveCleanup() call; the counted baseline shrinks.

* refactor(process): let native command markers read their supervisor's release

ProviderProcessSupervisor records an AnalysisCleanupRelease that a stop proving the process ended confirms. Native command and LLDB resolver cleanup markers bind to their supervisor's release, so a verified stop by either owner completes them, instead of each owner pushing resolveCleanup() onto the markers it tracks. Removes the last two resolveCleanup() calls outside the error base; the counted baseline is empty.

* chore(verify): forbid resolveCleanup() pushes outside the error base

Every production owner now confirms a release its errors read, so the counted baseline is empty. The guard becomes an outright ban with no baseline entry.

* test(native): complete native command and LLDB markers through a direct supervisor stop

* refactor(errors): complete an error's cleanup from its owner's record and every dependency (#10892)

* refactor(errors): complete an error's cleanup from its owner's record and every dependency

* refactor(errors): complete test errors through an owner release and delete resolveCleanup() (#10898)

* refactor(android): compose JADX and Apktool cleanup failures through one shared owner (#10900)

* refactor(android): compose JADX and Apktool cleanup failures through one shared owner

* refactor(application): compose service cleanup failures through the shared owner (#10902)

* refactor(sqlite): classify worker-reply read failures through the shared provider classifier (#10849)

* fix(pcap): report a refused tool launch as an unavailable tool

When a capture read's tool launch was refused with EPERM, pcap reported it
as unreadable tool output (AnalysisProtocolError) instead of an unavailable
tool. Pcap also never told "not executable" apart from "missing".

Add classifyLaunchFailure in src/process as the one owner of which spawn
errnos mean missing (ENOENT, ENOTDIR) and which mean present but not
executable (EACCES, EPERM, ENOEXEC, EISDIR). Pcap, apktool and gdb used three
different errno lists; all three now use it.

* refactor(process): classify owned command outcomes through one owner

Add ProviderOutcome, read once from the supervisor's facts, and classifyProviderOutcome, the one mapping from an outcome to a typed error. Migrate pcap and the shared artifact producer command failure (TShark, sigrok). Add the rea/provider-failure-classifier lint rule with a shrink-only baseline of per-site (cause: unknown) => AnalysisError functions.

* refactor(process): classify Java and adb launch refusals with the shared owner (#10598)

The Ghidra and JADX Java probes and adb's version probe now take
missing-versus-not-executable from classifyLaunchFailure instead of their
own errno checks. ELOOP and ENAMETOOLONG join the owner's missing set, so
adb keeps its existing classification and both Java probes now treat a
non-directory path component as a missing runtime.

* fix(pcap): preserve command launch attribution

* test(pcap): expect an unknown failing stage when a limited capture tool exits unsuccessfully

* refactor(process): classify owned command outcomes through one owner

Add ProviderOutcome, read once from the supervisor's facts, and classifyProviderOutcome, the one mapping from an outcome to a typed error. Migrate pcap and the shared artifact producer command failure (TShark, sigrok). Add the rea/provider-failure-classifier lint rule with a shrink-only baseline of per-site (cause: unknown) => AnalysisError functions.

* test(lint): cover baselined files in the provider-failure classifier rule

* fix(apktool): report the decoder JVM's heap exhaustion as memory, not a target refusal

The supervisor now watches for a managed runtime's out-of-memory report while output streams (JVM java.lang.OutOfMemoryError, V8 'JavaScript heap out of memory'), so a retention budget cannot hide it. ProviderOutcome gains a runtime-exhausted kind and classifyProviderOutcome maps it once. Apktool decode opts in and moves to the shared classifier; Ghidra's single-purpose output token becomes the JVM runtime option and its area matching reuses the shared table.

* fix(process): keep the runtime's out-of-memory report on the memory error it supports

* fix(pcap): attribute a limited capture's exit through the prlimit wrapper

* fix(process): keep a shortened out-of-memory line marked as shortened

* fix(process): identify the refused executable for every launcher shape

Name the executable on errnos Node throws synchronously from spawn
(ENOEXEC, ENOTDIR, ...) and read the Windows native launcher's
executable-launch constraint, so pcap attributes those refusals to the
selected tool or limiter instead of an unknown stage.

* test(pcap): prove synchronous launch naming with ENOTDIR on every POSIX host

* fix(apktool): advise the JVM limit that governs the reported memory area

* fix(ghidra): read the supervisor's parsed OOM area instead of reparsing lines

* fix(sqlite,capture): read V8 heap exhaustion from streamed output, not retained stderr (#10724)

The SQLite worker and the Node capture decoders now declare runtime: "v8", so the supervisor observes V8's heap report as output streams. Both classifiers read the runtime-exhausted outcome instead of matching retained stderr, so a report beyond the retention budget, or one whose process then overflowed its output budget, is still reported as memory exhaustion.

* refactor(evm): classify EVMole worker outcomes through the shared provider classifier (#10716)

* refactor(evm): classify EVMole worker outcomes through the shared provider classifier

The EVMole worker's cancellation, deadline, output budget and CPU/file-size signals now go through classifyProviderOutcome; only the prlimit launch refusal and the reserved EFBIG exit remain in its dialect. The worker is opted into V8 exhaustion scanning so a heap report names its fixed 256 MiB heap. providerOutcomeOf now reads a launch refusal only from a spawn errno, so a private-storage or carrier read errno is no longer reported as a limiter launch failure.

* fix(evm): keep a stream failure's exit status and update Linux worker expectations

* refactor(process): read launch refusals through the launcher identity owner

* refactor(wasm): classify WABT command outcomes through the shared provider classifier (#10727)

* refactor(wasm): classify WABT command outcomes through the shared provider classifier

WABT's cancellation, command deadline, output budget, and inherited-limit signals now go through classifyProviderOutcome; its dialect keeps only the configuration-probe and launch-refusal unavailability and the unclassified-exit output error. A tool ended by SIGXCPU/SIGXFSZ or the memory limit is now a resource constraint instead of an output error, and a refused launch in the producer phase is reported as unavailable. The shared timeout context gains the analysis setting that selected the deadline, and the output-limit context may name its exceeded budget.
…

---------

Co-authored-by: Ysopwnable <161784452+N0zoM1z0@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-11-16 10:59:18 +08:00
morluto 74dd9fc92c test(adapters): compare CLI and MCP Evidence for every CLI analysis operation (#11672)
* test(adapters): compare CLI and MCP Evidence for every CLI analysis operation

Run each official, workflow, native, artifact and managed operation that
has a CLI command through the CLI command tree and the MCP server over
one simulated provider, and require identical Evidence or failure. The
differences main has today are recorded per operation so the W2 invoke
path can only remove them.

Split bindDirectAnalysis(dependencies) out of createDirectAnalysis so the
CLI can run over injected session owners.

* test(adapters): give each parity case its own CLI and MCP connection
2026-11-15 21:56:55 -05:00
morluto 858ff73880 perf(tests): give MCP test clients the shared lazy schema validation (#10960)
* perf(tests): give MCP test clients the shared lazy schema validation

The MCP SDK client compiles every listed tool's output schema on its
first tool call. With the SDK's default Ajv options, REA's catalog costs
gigabytes of allocation per client (inspect_analysis_view alone: 1.7 s and
~500 MiB retained), so native MCP files peaked at 1-2 GiB per worker.

tests/fixtures/mcpSchemaValidation already owns lazy, reference-preserving
validation. Route the native MCP, process-boundary and shared harness
clients through it, share compiled validators across clients in a worker
(bounded LRU keyed by schema text), and add a counted architecture guard so
default-validation test clients cannot spread; the remaining copies are the
baseline.

* fix(guards): read call arguments past parentheses inside string literals

* perf(tests): give MCP boundary test clients the shared schema validation (#11022)

* perf(tests): give MCP boundary test clients the shared schema validation

* perf(tests): give the remaining MCP test clients the shared schema validation (#11025)

* fix(guards): read an MCP client's validator option from parsed code

The default-validation counter searched each new Client(...) call's text for jsonSchemaValidator, so an inline comment, a string literal or a client name naming the option hid a default client. Parse the source with @babel/parser and count NewExpressions of Client whose object-literal arguments have no jsonSchemaValidator property; a source that does not parse counts every textual call. The guard reads raw test sources, and the generated-child-script allowlist entry, which no longer counts as a call, is removed.

* fix(guards): read an MCP client's options only from a whole object-literal argument

The default-validation counter skipped files without the text
'new Client(', so a comment between the tokens hid a client, and it took
any depth-two key as an option, so 'enabled ? { jsonSchemaValidator } : {}'
passed while its false branch uses the SDK default. The skip now only
requires the name Client, and the validator counts only when it is a key
of an object literal that is a whole argument; anything else counts as the
default.
2026-11-15 21:53:53 -05:00
morluto f39e5e80db refactor(cli): own shell completion behind new CLI parity goldens (#11662)
* test(cli): capture CLI parity goldens before replacing incur

* test(cli): cover completion groups, aliases and hand-built enums in the parity goldens

* refactor(cli): own the shell completion protocol instead of incur's Completions (#11665)

* test(cli): keep root help goldens stable for longer names and let recording add cases
2026-11-15 21:52:17 -05:00
morluto d4c2be509e test: guard first-catalog, child CDP and queue regressions at the test deadline (#11675)
Three remaining wall-clock bounds become hang guards at the test's own
deadline, each with an observable outcome asserted:

- MCP startup: the pending initialize and tools/list phases get the largest
  budget Node can schedule, so only the first-catalog budget can end the
  wait before the test deadline. The elapsed bounds are deleted.
- Playwright storage child CDP reply: the elapsed bound is deleted. A
  missing command deadline leaves settle pending until the test deadline,
  and the asserted message names the value that was applied.
- Environment create explanation: the body has 1,000,000 values, so a
  quadratic queue walk runs past the test deadline (about 65 s). The
  linear walk takes under 200 ms. The elapsed bound is deleted.

Refs #10513
2026-11-15 21:51:05 -05:00
morluto 67b65b3540 fix(browser): keep the ZIP reader's fixed ceiling in a SAZ import refusal (#11670)
SAZ import reported every ZIP reader limit as its max_input_bytes budget.
A method-93 member whose Zstandard window exceeds the ZIP reader's fixed
128 MiB allowance was therefore refused as invalid_request out_of_range
with advice to raise max_input_bytes, which cannot admit it.

The SAZ decoder now keeps a ceiling the reader names: CaptureFormatError
carries it, the decoder child sends it on the reply wire as a
fixed-limit reply (allowance, value, unit, narrowing and the bounded
resource), and inspect_web_network_capture returns resource_constraint
with details.refusal for that ceiling. Only the metadata read budget the
decoder gave the reader still maps to the expanded-input budget.

A reader's fixed ceiling (ArtifactFixedLimit) now always states its value
and unit, so it can cross the process boundary intact. A refusal at a
producer-named fixed ceiling is built by fixedCeilingConstraint, which the
Playwright trace import also uses.

Fixes #11642
2026-11-15 21:41:25 -05:00
morluto b7fe131123 refactor(mcp): bind close_binary and reset_investigation through the shared table (#11647)
* refactor(mcp): bind close_binary and reset_investigation through the shared table

* refactor(mcp): bind provider operations through the shared table (#11649)

* refactor(mcp): bind provider operations through the shared table

* fix(mcp): bind a provider operation only to the contract of the same name

* refactor(mcp): bind objdump and Rizin commands through the shared table (#11653)

* refactor(mcp): bind objdump and Rizin commands through the shared table

* test(rizin): keep the pre-admission Rizin checks under the mutation sweep

* refactor(mcp): bind the desktop build tools through the shared table
2026-11-16 10:39:23 +08:00
morluto 07c5a54fee refactor(cli): import zod directly instead of through incur (#11655)
* refactor(cli): import zod directly instead of through incur

* fix(verify): count incur z imports from declarations, not comment or string text
2026-11-16 10:37:43 +08:00
morluto 832984c2cc fix(cli): describe a contract option whose description sits inside .optional() (#11661)
* fix(cli): describe a contract option whose description sits inside .optional()

zod does not carry a wrapped schema's description to its .optional()
wrapper. Since #10798 projected analyze_javascript_application from its
contract, source_paths is javaScriptSourcePathsSchema.optional(), so
'analyze --help' and 'analyze-javascript-application --help' listed
--source-path with no description and cliOptionAliases.test.ts fails on
main.

The contract option projection now reads the field's description through
its wrappers (describedThroughWrappers), once for every argv branch, in
place of three per-branch copies that read only the outer schema. The
opener and member option projections use the same reader.

* test(cli): assert the exact help text of a wrapped contract field and of a projection override
2026-11-16 10:35:14 +08:00
morluto 53fa95317a test: prove deadline sharing and early observation end without elapsed-time bounds (#11664)
The preflight digest case runs on a fake clock and asserts the timeout at exactly 200 ms. The fatal-protocol observation case uses a window longer than any test deadline, so waiting it out fails as a deadline instead of passing a 5 s bound.
2026-11-15 21:25:53 -05:00
morluto f060967789 feat(native): carry the value of each fixed structural decoding allowance (#11651)
* feat(native): carry the value of each fixed structural decoding allowance

BinaryResourceFailure and relocation-index refusals pass their maximum as limit and unit; the PE certificate decoder allowances drop the value from their names and pass it with a per-payload unit; the ELF note parser names its text ceiling with the enforced value. Kernel-module and unwind providers no longer claim a fixed text or structural allowance for a memory limit that names no ceiling (no parser produces one).

* fix(native): report a binary resource ceiling in the unit its producer counts

BinaryResourceFailure carries an optional unit; the ELF hardening entry budget says entries and the structural byte budgets say bytes. binaryResourceCeiling reports the unit only when the producer named it, so an entry count is no longer labelled as bytes. The ELF note test takes its expected ceiling from the fixture length instead of the message.
2026-11-15 21:25:13 -05:00
morluto 9953f0e46e fix(adb): classify adb launch failures through the shared launch owner (#11659)
* fix(adb): classify adb launch failures through the shared launch owner

The adb version probe and device commands read a launcher error with the
refusal-only table plus their own capacity check, so E2BIG became an
adb protocol failure and host capacity at the version probe became an
"unavailable" configuration. Both now read the launcher's failure with
launcherFailureOf and report host capacity and oversized launches
through classifyLaunchFailure; the selected-binary refusals keep adb's
own configuration reports.

The adb-specific capacity error, its remediation, the host_capacity
configuration code and the exported refusal-code helper are removed.

Refs #11614

* docs(adb): describe adb launch constraints as shared resource errors
2026-11-16 10:22:57 +08:00
morluto d0dded7d6e fix(artifacts): run hdiutil under owned supervision with a SIGTERM-first stop (#11632)
* fix(artifacts): run hdiutil under owned supervision with a SIGTERM-first stop

runOwnedCommand gains a per-command termination policy: before owned
cleanup kills a command's process group, it sends the command's own stop
signal to that group while the leader is unreaped and waits a bounded
grace for the leader to exit. NativeDmgArtifactReader runs every hdiutil
command through runOwnedCommand with SIGTERM and a 5 s grace, so an
interrupted attach can roll back as before. An hdiutil whose cleanup
fails is retained by the reader and retried on close, which reports it
until it is released. Raw command failures are interpreted once in
runChecked.

Refs #11619

* fix(artifacts): release hdiutil commands retained during DMG reader close

Detach and inventory commands that run during close can have their cleanup
fail too; close now retries retained hdiutil processes again after detach
and reports any that remain instead of resolving.

* fix(artifacts): keep hdiutil's exit status when its cleanup fails and budget each output stream

* fix(artifacts): admit no hdiutil command while an earlier one's cleanup is unresolved
2026-11-15 21:22:13 -05:00
morluto 162097b95a feat(gdb): name why close requires an explicit inferior disposition (#11652)
The pre-cleanup close refusal carries diagnostics.inferior_disposition_required (multiple-inferiors or no-durable-output-sink, the cleanup codes a forced stop would report) and live_inferiors, so callers and tests need not read its prose.
2026-11-16 10:00:19 +08:00
morluto 035d876d1f fix(lifecycle): supervise assetutil, keep cleanup-time interruptions, and cover browser-transport handshakes (#11502)
* test(lifecycle): cover browser-transport handshakes and cleanup retry

Close six provider lifecycle gaps: Electron CDP handshake, cleanup and retry; V8 inspector handshake; web-module-trace handshake and retry.

* test(lifecycle): assert retained cleanup finishes before the next acquisition

Hold the retried browser and Electron transport close and check that no replacement launch or connect starts until it completes. Track gated upgrade sockets in the fake CDP browser so close() destroys them.

* test(lifecycle): release held retry gates and join gated work on failure

Teardown for the resolver and Electron retry tests releases the held close and awaits the gated request even when an assertion fails while it is held.

* test(lifecycle): cover worker and tool launch, worker handshake, and decoder completion (#11512)

* test(lifecycle): cover worker and tool launch, worker handshake, and decoder completion

Close six provider lifecycle gaps: SQLite and JavaScript worker launch, JavaScript worker handshake, assetutil launch, and qlog and mitmproxy decoding through the owned decoder child.

* test(lifecycle): run the worker relaunch in a built project and check exits before close

The relaunch starts the built worker entry, so it moves to tests/boundary/javascript. The handshake tests now check that the stand-in child has exited when analyze() returns, before close() runs.

* test(lifecycle): cover in-process cancellation and PE certificate completion (#11524)

* test(lifecycle): cover in-process cancellation and PE certificate completion

Close six provider lifecycle gaps: caller cancellation for the managed, Harmony, JavaScript workflow, runtime reconciliation and V8 code cache providers, and PE certificate completion (an existing CLI/MCP regression). Two Apple application gaps remain (#11509).

* test(lifecycle): compare the whole retained V8 code cache observation

* test(javascript): hold the startup deadline clock until the stand-in worker is running

* test(javascript): restore real timers in a void teardown

* test(lifecycle): join the held Electron request on failure, refuse unmodeled Inspector commands, drop a wording assertion

* fix(apple): run assetutil under owned process supervision and retain unproven cleanup (#11586)

* fix(process): report a cancellation or deadline during refused final cleanup as that interruption

* fix(apple): prepare ownership before the assetutil deadline, keep per-stream allowances, join interrupted-cleanup tests

* fix(artifacts): classify assetutil launch failures through the shared launch owner

After #11615, owned-command launch failures are one launch-failed outcome
classified by the shared owner. The assetutil dialect keeps only its
unavailable report (launchUnavailable) and names its provider; host
capacity and E2BIG become resource constraints.

* fix(artifacts): keep a completed assetutil run's output on its cleanup-only report

When assetutil exits 0 but its process cleanup is unproven, the cleanup
error is the primary report. retainOwnedCommandCleanup now accepts the
completed run's captured output for such a caller, so CLI and MCP keep the
retained output beside the cleanup owner.
2026-11-15 20:58:47 -05:00
morluto e5cf021ca6 fix(process): classify GDB, Apktool and Java-probe launch failures through the shared launch owner (#11631)
* fix(process): classify GDB, Apktool and Java-probe launch failures through the shared launch owner

GDB session start, the Apktool version probe and the Jadx and Ghidra Java
probes read a launcher's error with launcherFailureOf and report host
capacity (EAGAIN, ENOMEM, EMFILE, ENFILE) and E2BIG through the shared
classifyLaunchFailure, keeping their own unavailable reports for refusals.
Only a launcher-raised errno counts as a launch failure, and a GDB launch
failure no longer retries the MI2 dialect. The refusal-only
classifyLaunchFailure helper is removed.

Refs #11614

* fix(process): name the environment as the E2BIG repair for fixed-argument launches

GDB, the Apktool version probe and the Jadx and Ghidra Java probes launch
fixed arguments, so selecting less input cannot shrink an oversized launch.
The shared launch report takes fixedArguments and then tells the caller to
shrink the environment REA runs with. Launch tests check structured facts
and guidance delivery instead of whole sentences.

* fix(process): report which launch input an E2BIG refusal leaves to shrink as a structured fact
2026-11-15 20:55:29 -05:00
morluto 0035d8edb3 feat(errors): name the fields a failure's recovery selects so the CLI renders their flags (#11644)
* feat(errors): name the fields a failure's recovery selects so the CLI renders their flags

AnalysisError gains related { parameters, operations } (contract fields and operation names), projected as details.related_parameters / related_operations and kept through the domain re-attribution helpers. The CLI projection adds details.related_cli_flags for its own command's fields. inspect_plist's missing-default failure and inspect_sqlite_database's column-evaluation failure name fields only in their text.

* test(sqlite): check MCP omits CLI flags on the failure that names related fields
2026-11-15 20:54:50 -05:00
morluto 2a31e064f3 feat(flutter): carry the value of each fixed Flutter allowance (#11641)
* feat(flutter): carry the value of each fixed Flutter allowance

Each Flutter fixed allowance now passes limit and unit, so details.refusal states the ceiling. The ABI inventory refusal names the ceiling that was exceeded: the ABI count or the library path bytes.

* fix(artifacts): name the ZIP decoder window ceiling in Flutter and trace refusals

A Zstandard member whose window exceeds the ZIP reader's fixed 128 MiB allowance now carries that ceiling as a typed fixed allowance (limit, unit, and the recompression narrowing). Flutter reported any ZIP reader limit as its APK metadata read allowance, and Playwright trace import as a recorder-chunk limit; both now keep the reader's ceiling.

* fix(flutter): report the APK expansion ceiling and keep JNI recompression advice

A ZIP reader limit with a reader-owned parameter (the size-derived expansion budget) is no longer reported as the Flutter metadata read allowance; Flutter exposes no control for it, so it is a fixed ceiling with the reader's effective value. The JNI library failure message carries the reader-named ceiling's narrowing, since that report has no refusal field.

* test(artifacts): compare decoder-window advice with its owning ceiling
2026-11-15 20:54:31 -05:00
morluto 2234a7db57 fix(mcp)!: require absolute local paths for pcap capture and debugger sessions (#11508)
* fix(mcp)!: require absolute local paths for pcap capture and debugger sessions

The MCP server's working directory is not the caller's, so sibling local
file contracts refuse a relative path. inspect_pcap_capture (path),
start_rizin_debug_session (path, inferior_tty) and start_gdb_session
(inferior_tty) admitted one and resolved it against the server's
directory. They now use absoluteLocalPathStringSchema.

The inspect-pcap-capture CLI command declares path as a caller path, so
a relative CLI argument is still resolved against the caller's working
directory before the contract sees it. debug-with-rizin and
debug-with-gdb call their managers directly and are unchanged.

BREAKING CHANGE: MCP calls to inspect_pcap_capture,
start_rizin_debug_session and start_gdb_session with a relative path or
inferior_tty are refused with invalid_request. Pass absolute paths.

Fixes #11079

* test(mcp): call the debugger session tools with absolute paths

* fix(cli): describe the inspect-pcap-capture path without the MCP absolute requirement

The CLI resolves a relative capture path, so its help keeps the
caller-path description instead of inheriting the contract's absolute
wording.

* test(pcap): expect the absolute-path format issue beside the empty-path bound
2026-11-16 09:48:52 +08:00
morluto ef031f639e feat(errors): carry the value of fixed allowances whose name embedded it (#11638)
* feat(errors): carry the value of fixed allowances whose name embedded it

Eleven fixed allowances (apktool output and version-output capture, the APK
signing report, release exposure input, external DEX representation, IPS fixed
format ceilings, native export rows and retained text, native plist input,
nesting and decode, PE import retained text) wrote their value into the
allowance name, so details.refusal had no limit or unit and the value existed
only in text. Each now passes limit and unit, the name no longer repeats the
value or 'no runtime override', and the derived remediation states both.

Refs #10508

* test(native): compare plist ceiling guidance with the refusal formatter
2026-11-16 09:44:10 +08:00
morluto 125612175a feat(pcap): carry the value of each fixed packet-tool allowance (#11643)
The packet-tool output capture, tshark JSON decoding and expansion, packet-inventory retained-fact, owned capture snapshot, and packet-field discovery record allowances now pass limit and unit, so details.refusal states the ceiling.
2026-11-16 09:42:07 +08:00
morluto 680c83f88c feat(errors): carry the value of Go, Rust, IL2CPP, sigrok, Electron fuse, CDP, projection and assetutil allowances (#11645)
Each of these fixed allowances now passes limit and unit, traced to the constant or parse refusal that produced it. The Electron fuse, CDP runtime retention and assetutil names no longer embed the value.
2026-11-16 09:40:58 +08:00
morluto 836ac2fea6 feat(artifacts): project fixed reader ceilings as typed refusals (#11634)
* feat(artifacts): project fixed reader ceilings as typed refusals

ArtifactReaderFailure('limit') carried fixed reader ceilings only as
remediation prose: artifactFixedLimitFailure and six Deno reader sites wrote
fixedAllowanceRemediation into the message, so CLI/MCP had no details.refusal
and the value was only in text. The failure now carries a typed fixedAllowance
(name, kind, limit, unit) that survives rebuilds (primaryFacts) and
translateArtifactFailure, and ArtifactOperationError projects it as a
fixed_ceiling refusal whose remediation derives from the allowance. The message
states only what exceeded the ceiling.

Refs #10508

* test(artifacts): keep a fixed reader ceiling's refusal through cleanup wrapping
2026-11-16 09:29:54 +08:00
morluto e94ca9f223 refactor(mcp): bind GDB and Cutter tools under external-mutation admission through the shared table (#11637)
* refactor(mcp): bind GDB and Cutter tools under external-mutation admission through the shared table

GDB start and status, and Cutter discovery, bind as values outside admission; gdb_console and cutter_command bind on the external-mutation evidence kind (gdb_console with delivery detail), and close_gdb_session on owned-cleanup. Tests build their servers through one debugger fixture, and the Rizin lifecycle regressions now run for all three debuggers.

* refactor(mcp): bind the session target, context and process capture tools through the shared table (#11640)

Value bindings can return a ReducibleValue (alternatives tried when the complete value overflows) or a DetailedFailure (facts delivered beside the error's details). A new query binding reads the analysis its required admission grants. open_binary and open_bundle_resources bind as values; get_navigation_context and inspect_address_context as queries; capture_process_scenario on the analyze kind through an application owner that records its Evidence and residual unknowns.
2026-11-16 09:28:35 +08:00
morluto 653a6d0359 test(snapshots): expect a snapshot binding mismatch to name targets by their persisted identity (#11639)
Since #11551 a snapshot persists its target as { identity, kind }, so binding_mismatch.target reports the identity's sha256 and selection there. Two tests still expected the flat target and fail on main.
2026-11-16 09:27:05 +08:00
morluto 093e41a568 refactor(mcp): bind Frida discovery and session tools through the shared table (#11621)
* refactor(mcp): bind Frida discovery and session tools through the shared table

* refactor(mcp): bind browser observation tools through the shared table (#11622)

* refactor(mcp): bind session record tools through the shared table (#11625)

* fix(mcp): record a completed provider-backed workflow's Evidence once before delivery

* refactor(mcp): bind Rizin debugger tools under external-mutation admission through the shared table (#11629)

* refactor(mcp): bind Rizin debugger tools under external-mutation admission through the shared table

* test(mcp): cancel a Rizin command during its retirement close, and route moved code to the reverse-tools lane
2026-11-16 09:14:20 +08:00
morluto 7b500d7833 fix(browser): report non-literal request URLs as unknown endpoints instead of dropping them (#11630)
* fix(browser): report non-literal request URLs as unknown endpoints instead of dropping them

* fix(browser): read XHR URLs from the proven request kind and bound view call sites
2026-11-16 09:12:44 +08:00
morluto 9738583ed8 fix(native,hopper): keep the descriptor a failed process identity read left open (#11633)
The LLDB tracer's target start-identity read and exit/signal identity checks,
and the Hopper client's bridge start-identity read and exit-identity check,
reduced an inspection-owner rejection to a message, so a descriptor whose close
failed was dropped. Each reader now keeps the identity unavailable (or the exit
unverified) and moves the rejection's descriptor owners into the session's
inspection scope, which trace/close retries; the reported failure names the
descriptor until it is released. The Hopper exit check no longer makes close()
reject.

Fixes #11602
2026-11-16 09:11:49 +08:00
morluto 9699ce9057 fix(frida)!: require an absolute script path over MCP (#11580)
* fix(frida)!: require an absolute script path over MCP

`load_frida_script` and `instrument_with_frida` admitted a relative script
`path`, which the MCP server read against its own working directory rather
than the caller's. Both contracts now admit only an absolute local path, as
the pcap and debugger contracts do. The `instrument-with-frida` CLI resolves a
relative `--script-file` against its working directory before the contract
sees it. `program` is unchanged: on a remote device it may name a path there.

BREAKING CHANGE: MCP `load_frida_script` and `instrument_with_frida` refuse a
relative script `path` with `invalid_request`; send an absolute path. CLI
`--script-file` still accepts a relative path, and Evidence `source_path` is
now the resolved absolute path.

Fixes #11506

* test(frida): assert a refused relative script path starts and loads nothing

* test(frida): verify the CLI fixture's script file spelling is relative

* test(frida): give the contract test an absolute script file

The load and instrument contracts admit only an absolute script path, so
the contract test's relative script.js no longer parsed. Use a resolved
path for the accepted inputs and refuse the relative spelling in its own
test.

* test(frida): drop the contract-level relative path check the MCP test covers
2026-11-16 08:54:51 +08:00
morluto fbea9e8778 fix(verify): read a TypeScript non-null assertion as completing its operand (#11596)
* fix(verify): read a TypeScript non-null assertion as completing its operand

A postfix ! after a completed operand leaves it complete, so a following / divides instead of opening a regular expression that hid the rest of the expression from the guards.

* fix(verify): read export default as the start of an expression
2026-11-16 08:52:06 +08:00
morluto 793ab582be refactor(desktop): give the agentenv provider name an entry file and name the build computer's controlled routes as a port (#11600)
* refactor(desktop): give the agentenv provider name its own entry file and move ComputerRelease into src/domain

* refactor(desktop): name the build computer's controlled routes and controller answer as a port (#11604)
2026-11-16 08:51:27 +08:00
morluto 2910fc0b13 fix(electron): refuse a wait longer than Node can schedule (#11606)
An Electron scenario wait accepted any non-negative duration_ms, and
Playwright's waitForTimeout schedules a delay above 2147483647 ms as 1 ms,
so a very long wait returned at once. Bound it at admission with the shared
Node timer maximum, as the browser scenario wait already does.

Refs #11601
2026-11-16 08:48:38 +08:00
morluto ad4f72dd68 build(scripts): type-check the shared verifier run, fixture, cleanup and source loader (#11609)
* build(scripts): type-check the shared verifier run, fixture, cleanup and source loader

The five scripts/lib modules join tsconfig.scripts.json (69 errors). Named types come from their existing declaration files. The source loader probes the runtime's TypeScript transform once and refuses a Node.js without transform mode (Node.js 26 removed it) with the pinned-runtime message instead of failing on the first owned source file.

* build(scripts): type-check the managed PE and function dossier fixtures (#11611)

* build(scripts): type-check the managed PE and function dossier fixtures

pe.mjs and function-dossier.mjs join tsconfig.scripts.json (68 errors). Named option types come from pe.d.mts; heap names are paired with their rows instead of read back by index; optional metadata tables are added to the row map directly; a placed body's RVA is computed by bodyRva, which refuses an unplaced body. Fixture bytes are unchanged.

* build(scripts): type-check the managed conformance entry point and support (#11612)

conformance.mjs, support.mjs and config.mjs join tsconfig.scripts.json (34 errors), completing the managed conformance closure. Result checks narrow with assert.ok, Evidence results are read through their owners' schemas, the native function subject is an EvidenceSubjectTarget, the comparison check no longer dereferences a right-only method's null left side, and the fixture callback returns the report instead of assigning outer variables.
2026-11-16 08:45:01 +08:00
morluto d6af08e4be test: name tests after the behavior they check, not a removed limit (#11618)
* test: name tests after the behavior they check, not a removed limit

Thirty-one test titles still said "beyond/above/after the former ..." for
limits that no longer exist. They now state what the test checks (the size,
count or depth it exercises, or the bound that still applies, such as
max_snapshot_bytes for the managed manifest snapshot). Test bodies are
unchanged.

Refs #10515

* test: say "more than" where a title counts only part of its fixture
2026-11-16 08:42:11 +08:00
morluto 34ec903786 feat(native): report which Mach-O slices a signature describes and their signing states (#11620)
* feat(native): report which Mach-O slices a signature describes and their signing states

inspect_signature gains slices: described (the codesign architecture the
identity, hashes, requirement, entitlements and runtime state describe),
architectures (every selectable slice), unmapped, checked and unsigned.
Partial coverage and unsigned slices were only limitation prose, so the
native inventory goldens matched sentences. Slices checked before a failed
per-slice read keep their states in the partial observation.

Refs #10508

* fix(native): name a thin slice by codesign's architecture and qualify unknown slice states

A thin file's admitted architecture is a CPU family, so slices.described
reported a thin arm64e file as arm64. It now takes the architecture codesign
names for a thin file (null when it names none) and keeps the selector for a
universal one. The described slice's state is the top-level signed; only
other unchecked slices are unknown, and architectures comes from the admitted
universal header.
2026-11-16 08:40:43 +08:00
morluto d4a7c17b38 feat(errors): project fixed native and producer input allowances as typed refusals (#11623)
* feat(errors): project fixed native and producer input allowances as typed refusals

fixedNativeAllowance wrote its allowance and value only into the remediation
sentence, and fixedProducerInputAllowance put the byte count into the
allowance name. Both now build fixedAllowanceOptions with limit and unit, so
the native command output capture, PE resource metadata, Ghidra
installation-properties and artifact analysis input refusals project
details.refusal {kind: fixed_ceiling, allowance_kind, limit, unit}. The
fixed-ceiling remediation renders the value from those facts for every
ceiling that carries them.

Refs #10508

* test(errors): check the rendered allowance value, not its sentence
2026-11-16 08:39:18 +08:00