Commit Graph
3162 Commits
Author SHA1 Message Date
morluto 3b272dc8e2 style(apple): normalize class parser module formatting 2026-10-15 03:08:26 +08:00
morluto 5bea037552 refactor(apple): remove unused class identity re-export 2026-10-15 03:07:39 +08:00
morluto 2c36f8d8e5 style(test): format coverage regression fixtures 2026-10-15 03:06:59 +08:00
morluto 054be92874 docs: align workflow recovery and current Evidence contracts 2026-10-15 03:06:26 +08:00
morluto 67f4b79b32 fix(browser)!: expose capture budgets and queue admitted content reads
BREAKING CHANGE: Browser capture results require resolved limits, event families, and omission facts. Caller-selected budgets replace hidden retention ceilings; admitted content reads queue behind bounded concurrency.
2026-10-15 03:06:26 +08:00
morluto b01db67586 test(hopper): verify external exit recovery and retained ownership 2026-10-15 03:06:26 +08:00
morluto 3c4bd35180 fix(jeb): probe live operations with owned discovery deadlines 2026-10-15 03:06:26 +08:00
morluto b647e4a4e8 fix(cli): preserve Cutter choices and reject invalid runtime arguments 2026-10-15 03:06:26 +08:00
morluto 7f2f0b78c4 fix(ghidra): preserve MIPS overrides and filter entry metadata 2026-10-15 03:06:26 +08:00
morluto 61141032f7 fix(windows): explain missing native artifacts and verify output drain 2026-10-15 03:06:26 +08:00
morluto 303dc2aa73 fix(setup): own Hopper replacement staging and committed outcomes 2026-10-15 03:06:26 +08:00
morluto 98c3c527d8 fix(apple): distinguish JVM resources and UIKit event connections 2026-10-15 03:06:26 +08:00
morluto 5f986d4869 fix(process): attribute finite trace omissions and cleanup states 2026-10-15 03:06:26 +08:00
morluto b9539e9c1c fix(functions): preserve coverage diagnostics separately from blockers 2026-10-15 03:06:26 +08:00
morluto 92e17ccfce fix(javascript): preserve module identity and boundary fields 2026-10-15 03:06:26 +08:00
morluto 9392590f2f fix(artifacts)!: authenticate complete inspection and summary inputs
BREAKING CHANGE: Artifact inventories require scoped coverage. Obsolete coverage_status and prose-based coverage inference are removed. Inspection carriers authenticate every retained inventory substep.
2026-10-15 03:06:26 +08:00
morluto ef0c8005bc refactor(evidence)!: require current coverage commitments
BREAKING CHANGE: Evidence and analysis snapshots require explicit nullable relationship coverage and include it in their identity commitments. Historical records without the field are rejected; saved files are not migrated.
2026-10-15 03:06:26 +08:00
鸡哥 75104040df fix(javascript): track public CommonJS export receiver lifetimes (#3567)
* fix(javascript): track public CommonJS export receiver lifetimes

* fix(javascript): retain exports across uncertain identity effects
2026-10-15 02:46:14 +08:00
357996cc89 test(perf): measure genuine SAZ selection and retained recovery (#3584)
* test(perf): pin genuine Fiddler captures and raw HTTP oracles

Pin two public Microsoft SAZ captures with 13 and 1311 sessions, immutable source/digest/member coverage and predeclared caller-ordered transaction questions. Independently preserve stored HTTP headers, binary/empty bodies and archive coordinates; admit only known corpus bytes before ZIP allocation. Five reader regressions and both genuine selected-record checks pass. No production or performance-gain claim.

Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>

* test(perf): measure SAZ selection and retained recovery workflows

Add authenticated complete/ordinal/retained-page SDK experiments, fixed real-Codex prompts and independent cited-byte grading. Preserve actual failed RPCs and exact retained-reference recovery when complete MCP delivery exceeds its receive buffer. Verify selected CLI/MCP facts, parameters and semantic Evidence identity. Keep a genuine partial-message admission-negative probe. Focused 12 tests and 10 actual-Evidence grading cases pass; no formal model or gain claim yet.

Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>

* test(perf): verify resolved SAZ Evidence selection parameters

Distinguish omitted request ordinals from the complete resolved ordinal scope recorded by WebNetworkCaptureService. Share the expected sanitized Evidence representation between SDK checks and model grading; preserve original caller arguments. A focused regression and both saved genuine baseline RPCs confirm the correction. Keep original collector failures and costs unchanged; freeze a new driver before repeated measurements.

Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>

* docs(perf): retain mixed SAZ whole-task costs and failure boundaries

Record both SDK batches, twelve corrected workflows, all four actual Codex attempts, independent byte equivalence and verified payload-phase CPU/memory. Distinguish successful-analysis MCP transport limits from real provider timeouts and negative partial-message admission. Three model questions complete; the only completed pair saves 1.6 percent input but takes 2 percent longer, with no consistent benefit. Keep production APIs/defaults/cache unchanged and preserve all deviations, failed attempts and unknowns.

Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>

---------

Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-10-15 02:45:44 +08:00
e79e014f23 fix(browser): preserve passive network inventory uncertainty
Preserve uncertainty in offline passive network comparisons when request
identity, metadata, or per-kind inventory coverage is incomplete. Retain
proven HTTP/WebSocket changes without using missing information to infer
membership, duplicate counts, or equality.

Compare redirect resource types only where both observations retain them.
Keep metadata completeness with its owner, stream HTTP content digests,
and require full coverage for multiplicity and WebSocket stream inventories.
Preserve independently known metadata changes and full-inventory counts.

Keep the original JavaScript configuration-escape, JSON, cancellation,
SAZ and schema-documentation fixes compatible with current main. Split
public CLI comparisons into isolated test cases without loosening timeouts.
Validate actual Chromium/HTTP captures through CLI and real stdio SDK,
with explicit limits and unchanged source bytes.

Co-authored-by: To3akaRin <131595336+To3akaRin@users.noreply.github.com>
Co-authored-by: N0zoM1z0 <161784452+N0zoM1z0@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-10-15 02:28:46 +08:00
morluto b8510ee00b test(mcp): remove cache and event order dependence from fault probes
Inject IDA diagnostic capacity faults per connection so prior module imports cannot bypass them. Use the CDP fixture event barrier before asserting complete body shapes, and document profile and renderer verification.
2026-10-15 01:41:31 +08:00
morluto 36615d9165 fix(mcp): honor Frida partial evidence retention acknowledgments
Use shared recorded-partial delivery on start, load and instrument errors. Return a retained reference only after a successful write; preserve the original inline partial when retention fails or is unavailable.
2026-10-15 01:41:31 +08:00
morluto c5fb86d194 fix(mcp): bind catalog identity to output schema profiles
Include the selected output advertisement profile while retaining canonical output validation in the identity. Reuse the SDK output projection and select doctor expectations from both profiles. Correct the compact input error documentation.
2026-10-15 01:41:31 +08:00
morluto 0df0e4f53e fix(contracts): honor options outside the schema projection cache
Bypass target-only memoization when Standard JSON Schema library options are supplied. Verify shared and inline projections remain independent in either call order.
2026-10-15 01:41:31 +08:00
morluto 204b9f3ac1 fix(contracts): preserve array guidance across MCP schema profiles
Keep item schemas when compact inputs reduce nested structure, expose literal tuple elements to items-first renderers, and preserve SQLite table-selection guidance across compaction. Verify catalog-wide client constraints and canonical rejection before dispatch. Document the stock Pi renderer size limit and caller-selected recovery.
2026-10-15 01:41:31 +08:00
Dery Ferdika a732c859dd fix(contracts): reject NUL bytes in Frida, Rizin, and GDB local path inputs (#3575)
Route the caller-selected local path fields in load_frida_script,
instrument_with_frida, start_frida_session, start_rizin_debug_session,
and start_gdb_session through the shared localPathStringSchema so a NUL
byte is refused at the advertised contract boundary instead of reaching
the filesystem or a provider subprocess and surfacing as a raw error.

Extend the OS-invalid-local-input acceptance harness to the tools that
were missing from it.
2026-10-15 01:17:54 +08:00
morluto 10aa74dfed fix(mcp): preserve tuple constraints in advertised schemas
Declare fixed tuple tails and lengths at the canonical boundary so client array validation accepts every tool. Preserve nonempty candidate constraints in managed and application comparisons. Add catalog-wide client checks and parser/schema parity coverage for both dialects and projection directions.

Fixes #3576
2026-10-15 01:12:46 +08:00
93e75bc8fe fix(setup): retain Hermes YAML scalar source during updates
Retain source nodes for unchanged scalar values and string keys when
replacing a Hermes registration. Discarding quoted YAML 1.2 keys could
make Hermes' C reader reject duplicate boolean keys and switch to the
pure reader, changing the interpretation of unrelated settings.

Use linear mapping lookup, quote newly created YAML 1.2 strings for
YAML 1.1 compatibility, and preserve direct numeric edit formatting.
Extend actual CLI coverage for both existing reader-selection directions;
verify strings, exact unrelated numeric/timestamp values, backups and
idempotence with the actual Hermes Python loader.

Follow-up to #3102 and main's independent Hermes scalar fix.

Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-10-15 00:57:57 +08:00
1579b6865e test(electron): isolate active-hook fixture resolution (#3565)
Stage the real active hook and boundary helper next to a fixture-owned CommonJS Electron package so preload and exercise resolve the same module. Remove NODE_PATH reliance, retain capacity/preload/fatal-exit coverage, and prove isolation against a conflicting ancestor package.

Fixes #3473

Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-10-15 00:44:48 +08:00
569ef4159d feat(website): add contributors and complete the Greek edition (#3520)
Add an equal, searchable directory of 214 contributors and translate its shared
navigation and copy. Complete the Greek website edition (1728/1728 messages),
preserving source examples, protected commands and evidence.

Publish refreshed contributor snapshots atomically and cover partial writes,
denied replacement and staging cleanup with failure regressions.

Validation: complete website checks and 16 maintenance tests, strict Greek
coverage, desktop/mobile and no-JavaScript journeys, selected PR CI and Greptile
follow-up. Codex automated review was unavailable because its quota was exhausted.
No analysis-provider behavior or production website publication is included.

Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
Assisted-by: Suika (gensokyo-skills:suika-scatter-gather-planning)
Assisted-by: Kogasa (gensokyo-skills:kogasa-surprise-testing)
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-10-14 16:21:24 +00:00
0f623732bd fix(ci): require portable inventory reports on selected runs (#3549)
Run complete portable-inventory validation whenever code tests are selected and the job is not cancelled. Missing timing reports now fail rather than bypassing the existing validator. Preserve partial diagnostics and successful-history ownership.

Verified the baseline regression, missing/present report and cancellation/selection controls, actual Vitest blob/report failure paths, static checks, actionlint and test discovery.

Fixes #3542

Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-10-15 00:16:09 +08:00
044fb9127a fix(hopper): skip non-file launcher candidates
Require automatically discovered Hopper launchers to be executable regular
files so POSIX directory search permission cannot hide a later valid
application bundle. Preserve file symlinks, explicit caller configuration,
and the shared candidate ordering.

Cover canonical and renamed user bundles through the production filesystem
predicate; respect existing canonical or renamed system installations in
the isolated fixtures. Clarify launcher eligibility in the installation guide.

Follow-up to #3085 and #3054 after their shared path discovery landed on main.

Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
2026-10-15 00:09:42 +08:00
1438422dfe fix(managed): preserve enclosing identity in member comparisons (#3079)
Keep same-named nested members paired by their source-owned enclosing identity, including selected inventories that omit the parent TypeDefs. Share the cached, bounded nominal type-name digest owner with signature resolution; preserve current unknown identity through matching, remapping and absence instead of repairing it with display names or historical compatibility.

Verification: 457 related tests passed after preserving TypeDef visibility/parent applicability and coherent fixtures; the complete static/architecture gate and real pinned-SDK MCP nested/official .NET DLL workflows passed after a sigrok-experiment-only main refresh. Documentation checks passed after the compatibility fix. Full CoreLib inspection retains the pre-existing explicit representation limit. No target assembly execution; local verification used two CPUs and an 8 GiB RSS guard.

Co-authored-by: 8rulerstar <190670431+8rulerstar@users.noreply.github.com>
Co-authored-by: N0zoM1z0 <161784452+N0zoM1z0@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
2026-10-14 23:20:07 +08:00
YsopwnableandKaoru0822 eb44921c4c fix(ci): invalidate GitHub entrypoint and package-scope caches (#3522)
Track GitHub package scopes and Knip workflow/action entrypoints in static-check cache keys. Preserve the upstream action-code ownership fix and restrict TypeScript to its actual typed inventory.

Real cold/warm regression tests retain cache reuse for unrelated inputs while rejecting changed source, declarations, package scopes, manifests and reachable module dependencies. Existing lint inventory and report transport coverage remain unchanged.

Fixes #3510

Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
2026-10-14 23:15:04 +08:00
ca1b2f8608 test(perf): measure real sigrok annotation selection workflows
Pin two public sigrok recordings and independently verify raw waveform bytes/sample locations through real CLI/MCP selection workflows. Record all eight actual Codex task costs, including a fixed-plan deviation and mixed whole-task token/time outcomes. Existing annotation controls reduce reply bytes; the cohort does not justify a new production cache/default/API change.

Harden corpus identity, UART data-bit coverage, I2C interval ownership and preallocation ZIP/metadata admission. Focused tests and original Evidence/transcript replays retain observed facts, outcomes, provenance and resource limitations.

Fixes #3131.

Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
2026-10-14 22:56:25 +08:00
Ysopwnable 0f55f247cd fix(ghidra): preserve PE PDB coverage facts (#2933)
Preserve reported Program PDB identity, explicit omissions and nullable load state, retaining declared CodeView facts when equivalence is unknown. Bound JSON locator expansion and advance PE/workflow replay revisions without discarding historical Evidence.

Main already contains the shared canonical workflow collector. Integrate its unknown-PDB repair and fix the CLI fixture for the new artifact-session dependency. No unrelated pcap repair is included.

Validation: 309 focused tests across 15 files pass (2 Windows-only skips), runtime build and check:fast/pre-push pass. Real Windows Ghidra 12.1.4 PUBLIC/JDK 21.0.12.1 uses the public where.exe/PDB pair; 17 relevant production files match verified ca24903f. MCP cold/warm normalized results match and the backward graph has 3 successful nodes. Inputs are unchanged and no owned Java/Node processes remain. CLI inspection keeps its earlier exact-source scope. NB10 omission uses regression fixtures, without a real NB10 provider claim.

Tests keep 2-core affinity and 512 MiB Node/Java heaps. Retain the 600s analysis and 30s initialization timeout observations alongside the successful test-only extended-budget run; no product-default or cross-revision speedup claim.

Fixes #2259
2026-10-14 22:25:05 +08:00
morluto 6060de329d fix(release): rehearse qualified packages before publication 2026-10-14 21:51:19 +08:00
morluto c0aa2e32b3 fix(ci): include GitHub action code in static cache inputs
Invalidate typecheck, lint, built lint and Knip when action implementations or declarations change. This closes the real checker inventory gap detected by staticCheckReports on main.
2026-10-14 21:42:54 +08:00
eab789aed7 fix(managed): compare nominal signatures across metadata builds (#3059)
Compare managed member signatures using bounded, framed nominal type identities rather than build-local metadata row numbers. Preserve unresolved and historical evidence semantics, and share TypeRef, nesting and member-list facts at their producing boundaries.

Validation: current-head CI and Greptile pass; all 435 related managed tests pass. Genuine Microsoft .NET DLL CLI comparison and bounded SDK MCP paging pass without modifying targets. A local CLI filter deadline also fails on clean main; no unrelated timeout or product change is included.

Co-authored-by: JayOfTheKeyboard <140487034+JayOfTheKeyboard@users.noreply.github.com>
Co-authored-by: N0zoM1z0 <161784452+N0zoM1z0@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
2026-10-14 21:16:48 +08:00
morluto 790657dc0d fix(release): recover verified partial registry publications
Validate original exact-source qualification, release tag, retained package provenance and npm integrity before the corrected public-package canary. Publish only the existing release's MCP metadata through a shared idempotent publisher; never recreate tags or npm versions.
2026-10-14 21:15:34 +08:00
morluto 0b4fd11c8c fix(release): install the npm canary before its MCP handshake
Complete the isolated published installation before starting the protocol deadline, disable dependency lifecycle scripts consistently with package verification, and reuse that installation and cache for update validation.
2026-10-14 21:12:22 +08:00
morluto c99c6a89d6 fix(release): restore qualified runtime before the npm canary
The published-package verifier imports contracts from dist in the release checkout. Restore the exact qualification run's runtime artifact before publication so those dependencies are available without rebuilding the package.
2026-10-14 21:05:20 +08:00
3b79af28e8 chore(main): release rea-agents 6.5.0 (#1693)
* chore(main): release rea-agents 6.5.0

* docs(release): finalize contributor-aware 6.5.0 notes

* docs(release): retain verifier maintenance only in provenance

* style(release): format the contribution inventory

* docs(release): credit Windows fixes

* fix(comparison): recheck historical calls-only changes (#2223)

* fix(comparison): preserve incomplete function call evidence

Keep missing, unknown, partial and unresolved call relationships unknown in function comparisons and reconstruction claims. Aggregate IDA caller and callee coverage without overwriting uncertainty.

Validate with 144 focused tests, typecheck/lint, docs checks and test discovery. Real providers were not verified; the full headless IDA session suite requires unavailable Windows native controls.

* fix(comparison): recheck historical calls-only changes

---------

Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>

* fix(ci): isolate Linux Hopper process fixture inventory

* docs(release): credit selected-null and client setup changes

* fix(release): refresh 6.5.0 notes and align installation examples

* docs(release): clarify 6.5.0 migrations and recent fixes

* docs(release): cover omitted capabilities and refresh 6.5.0 provenance

* docs(release): pin final CI corrections in 6.5.0 inventory

* docs(release): include mutation scheduling in 6.5.0 provenance

* docs(release): include cache policy correction in 6.5.0 provenance

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
Co-authored-by: Codex <codex@users.noreply.github.com>
Co-authored-by: Lucien <164857476+dakjdakd@users.noreply.github.com>
rea-agents-6.5.0
2026-10-14 20:31:45 +08:00
morluto cbc0e97c29 test(ci): bind npm cache IDs to output consumers
Remove the blanket caller ID requirement that rejected the standalone mutation workflow. Keep producer bindings where static cache authorization and package verification consume action outputs.
2026-10-14 20:10:17 +08:00
morluto 92a0edf187 perf(ci): move full mutation sweeps off the release path
Keep fresh coverage and both real mutation preflights required in every CI tier. Run complete canonical sweeps daily or manually in isolated jobs with distinct reports, retained failure diagnostics, and explicit report-presence checks.

Preserve regression inventories, release qualification and mutation settings; remove the superseded duplicate gate test. Related: #3281.
2026-10-14 19:49:08 +08:00
N0zoM1z0andKaoru0822 1b46398795 perf(ci): preflight mutation configurations during fast feedback
Keep full qualification and standalone mutation execution intact, retain distinct reports, and preserve fail-closed quality gates.

Related: #3281

Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
2026-10-14 19:48:30 +08:00
morluto f75b1a416b test(browser): retain bounded diagnostics for scenario policy failures 2026-10-14 19:24:14 +08:00
morluto 0686b07546 test(doctor): retain discovered Hopper paths despite version failures 2026-10-14 19:23:11 +08:00
morluto c2b79aa462 test(doctor): separate Hopper availability and version assertions 2026-10-14 19:10:22 +08:00
morluto efe0923914 fix(release): exclude Codex automation from contributor thanks 2026-10-14 19:06:36 +08:00