Finish every remaining iterator from inner to outer, retaining cleanup observations without replacing the original traversal failure. A real nested ASAR regression reproduces the previous integrity-error masking.
Keep directory handles under explicit close ownership and retain ZIP and Mach-O admission owners when initialization or cleanup fails. Preserve cancellation and primary failure details, and avoid false successful release after native handle invalidation.
Validation: filesystem ownership, FIFO, archive and reconstruction boundary checks; static and architecture checks; real macOS CLI/MCP bundle workflows. Windows native behavior was not verified.
Require an ArtifactResourceScope for web script exports and share its lifetime with CLI or MCP shutdown. Retain setup-failed output trees in firmware and Wakaru providers. Preserve completed publications when descriptor cleanup fails, returning the full result after a successful retry or as a partial observation while cleanup remains unresolved.
Use bounded positional reads for unpacked ASAR members and borrowed Mach-O slices so destroying a stream cannot bypass its descriptor owner. Retain uncertain ASAR closes and preserve borrowed sources after cancellation.
BREAKING CHANGE: Injected ASAR unpacked handles expose fd and read instead of createReadStream.
Keep rejected closes unresolved after descriptor invalidation and share concurrent close attempts. Preserve primary admission failures and transfer the exact failed handle owner into inventory scopes.
BREAKING CHANGE: ArtifactResourceScope file-handle owners require OwnedFileHandle. Root inventory sources retain their single cleanup owner.
Return an unverified cleanup failure before invoking numeric process-tree termination when owned lineage is unavailable. Preserve the partial capture and existing cleanup diagnostics.
Exercise the provider capture boundary with stable application-state observations; assert no Windows cleanup call and retained partial Evidence. Source tests pass; native Windows behavior was not verified.
Keep failed reader, root descriptor, and output-tree cleanup with the caller. Drain admitted work before close, retry the same retained owners, and block new resource work while cleanup is incomplete. Preserve primary failures and complete inventory observations.
Thread ownership through JavaScript reconstruction, the CLI's one-shot scope, MCP server shutdown, verifier scripts, and test fixtures. Remove redundant ownership flags and the unused validated test wrapper.
Validate with 850 passing focused tests across 90 suites (two skipped), targeted classification and cleanup retries, static and unused-code checks, and real macOS CLI/MCP directory, ZIP, and DMG workflows. Native Windows and deep-analysis providers were not verified.
BREAKING CHANGE: Direct artifact inventory/extraction and JavaScript analysis APIs require an explicit caller-owned ArtifactResourceScope. CLI and MCP input contracts are unchanged.
Build each side's digest-to-Evidence index once instead of scanning native observations for every comparison item. Retain every exact match and avoid linking subjectless Evidence to nodes lacking content identity.
Comparison and relationship regressions pass. Isolated alternating unprofiled runs at 2,000 nodes and observations per side have conventional medians 379.924 ms before and 246.466 ms after; the complete 4,000-item outputs match byte-for-byte. Timings are bounded workload evidence, not whole-suite performance.
Flameox profile: flameox://evidence/7dca714833dc3d254a75704670a81de656b4b31d4b77a72236eecb82459b3d86
Use the shared nonblocking, no-follow regular-file opener at the source-read boundary and report a nonregular replacement as changed. Keep descriptor identity and extent checks.
The compiled subprocess regression timed out on the baseline and completes without a FIFO writer after this change. Reference reader failure, capacity, cancellation, and import checks pass.
Inspect each entry without following links before applying directory-only patterns. Retain the winning rule and original entry path while pruning excluded directories before reading their children.
Validate directory exclusions, negation, symlinks, explicit secret rules, and compiled CLI provenance alongside reference reader failure and cancellation coverage.
* feat(ghidra): admit set_address_name and set_addresses_names
Ghidra sessions rejected the existing naming tools with "operation is not declared by this provider", so agents had to use annotate_native_function, which returns a complete function dossier per rename and cannot name data addresses.
The Java bridge now names a local function entry (namespace-aware, like annotate_native_function) or creates/renames the primary label at any other mapped address, verifying readback. set_addresses_names applies each address in its own transaction and reports per-address booleans. Both are mutating capabilities: excluded from read-only (Windows P0) handshakes and from snapshot replay.
* fix(ghidra): normalize qualified symbol names and reject empty leaves
Share namespace-aware leaf normalization across functions and primary labels. Invalid qualified leaves now return false for naming operations without marking the provider unavailable; function annotations retain typed invalid-input errors.
Extend real CLI/MCP namespace verification with namespaced globals, readback replay, failed-name rollback, mixed batches, and capability checks.
Validation: 75 focused tests and npm run check pass. Ghidra 12.1.4/JDK 21 namespace verification passes on macOS arm64 with unchanged source bytes and complete cleanup. The full host lane passes this workflow before the existing entry-alias fixture fails because the compiler places its interior label at the function entry.
---------
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
Make inventory reader construction synchronous and DMG attachment lazy, so the inventory registers its reader before native operations start. Share initialization, reject work after close, wait for an in-flight attach before cleanup, and preserve retryable close state.
Cover failed-attach inventory cleanup, concurrent initialization, close-before-use, and close-during-attach alongside existing ownership, cancellation, and integrity cases. Session-level retention after inventory cleanup failure remains a separate concern.
Use the descriptor size to bound each allocation, with a one-byte EOF probe, and stop on observed growth. Retain short-read, cancellation, and final identity validation.
A 2,001-file benchmark requests 4,200,307 bytes instead of 266,469,376 with the same captured-byte digest. Flameox shows filesystem waits dominate, so this is an allocation reduction without a CPU speedup claim.
Record the exact matching pattern and distinguish project, default, caller, and configured-sensitive exclusions without changing ignore ordering. Include that provenance in graph commitments and validate the contract through compiled CLI and MCP consumers.
BREAKING CHANGE: Pattern-based reference graph exclusions require pattern and distinguish project-ignored and default-ignored reasons. Re-import saved graphs that omit matching patterns.
Pass typed workflow results directly to createEvidence, which validates generic JSON and takes the owned snapshot. Remove the extra generic parse and full result clone from all five workflow helpers; domain producers retain their specific schema validation.
Flameox confirmed JSON validation and copying costs. Alternating warm runs creating Evidence for a valid 2,001-node trace reduced median boundary time from 15.94 ms to 11.20 ms over 120 samples per variant, with identical output digests and independent mutable-input snapshots. This does not measure end-to-end workflow latency.
Validation: 39 focused CLI, SDK MCP and filesystem workflow tests and 17 Evidence tests passed; build, static and documentation checks passed. No deep analysis providers were exercised for this change.
Reject Mach-O and ZIP entries whose metadata no longer matches the reader inventory, preventing a sibling adapter key from selecting bytes under another entry. Reuse the existing ASAR metadata snapshot and equality rules and read Mach-O ranges from owned records.
Move the real system-lipo test to the provider boundary suite, release its reader, and derive slice expectations from the host FAT table.
Validation: 68 focused artifact, ASAR and MCP tests plus ZIP cancellation passed. Real macOS lipo was exercised against /bin/ls; Ghidra, IDA and other deep providers were not run. Static, docs, architecture and test discovery checks passed.
Share root filesystem failure classification between import preflight and traversal. Missing roots remain invalid selections; permission and I/O errors retain their reason and reach the CLI as execution failures. Preserve cancellation priority and expose unexpected exceptions.
Validation: compiled CLI permission regression, direct reader error details, import/traversal and unreadable-directory boundaries passed. Windows-only cases were skipped on macOS; static, architecture and documentation checks passed.
Avoid repeatedly canonicalizing scalar seeds for every literal node. Valid literal metadata without a scalar value now remains unmatched instead of throwing.
Flameox identified canonicalization as the hotspot. Alternating warm query benchmarks over 2,002 nodes reduced the 35.5 KB seed median from 36.34 ms to 0.158 ms; a short seed improved from 0.304 ms to 0.128 ms with identical result digests. These measurements cover the query function, not end-to-end analysis.
Validation: 38 domain query/traversal tests and focused compiled CLI/MCP consumers; static, architecture and documentation checks passed.
Own unpacked FileHandles until closure succeeds, deduplicate in-flight closure, and retain cleanup resources for retry without replacing the original read or identity failure. Attempt handle and snapshot cleanup independently and retain failed snapshots.
Consolidate unpacked admission, identity checking, cancellation and stream setup, removing the redundant same-handle stat. Cover stat and identity failures, post-stat cancellation, combined cleanup failure/retry, and stream completion/cancellation with real file descriptors.
Validation: 32 ASAR boundary tests, artifact extraction and SDK MCP consumers, check:fast. Native Windows descriptor inspection and real-provider workflows were not verified.
Use resolved seeds directly, retain one list of relevant unknowns, and stop sorting adjacency used only for complete reachability. Preserve canonical graph-array output order and candidate coverage scans.
Validation: 37 query and traversal tests, public JavaScript reconstruction and invalid-Evidence consumers, check:fast. No real-browser or deep-provider workflow verified.
Project authenticated inline Evidence failures at source.evidence while retaining their reasons and messages. Supply the matching request root to input issue projection so nested values remain identifiable.
Validation: both service entrypoints and SDK MCP integrity regressions, advertised-schema consumer suite, check:fast and architecture guards. No real-provider verification.
Count already-validated records and unknown revisions without reimporting, sealing, and exporting a temporary empty ledger. Keep canonical integrity checks at the file boundary and verify counts and malformed bundles through the compiled CLI.
Validation: Evidence file and CLI consumer suites; check:fast; Flameox CPU profiles and alternating 3.8 MB import benchmark (315.8 ms to 102.5 ms warm median, identical result digests). No real-provider workflow changed or verified.
Retain the original owned-command cleanup failure and one captured-output copy while preserving cleanup resources, truncation and retry semantics.
Verify socket release after cancelled CDP cleanup while retaining the independent detach regression.
Validation: 167 tests across 13 files, all eight local gates, real small Chrome/EVM CLI and pinned-SDK MCP controls, and 20/20 exact-head CI checks.
Derive needed artifact IDs directly from selected occurrences. Keep the node lookup local to selection, drop the unread occurrence map and collector helpers, and pass only manifest and selected integrity metadata to materialization.
Validation: existing extraction completeness, identity, unsupported-format, packed/unpacked ASAR, and public MCP workflows pass. Nested archive selection and equal-content occurrence coverage remain intact. This is a structural simplification; no throughput improvement is claimed.
Remove the empty application-node index from incremental projection state. The completed application graph now owns the only binding pass, and indexing returns the map directly instead of a wrapper.
Validation: 76 builder/composition tests and 19 filesystem, CLI-output, and SDK MCP trace boundary tests pass. Identity budgets and streaming/full-projection parity remain covered; no throughput improvement is claimed.
Validate columns × (rows + scrollback) against a 1,000,000-cell xterm working-buffer budget, including every resize with the selected scrollback. Document the budget in the advertised MCP input and process-capture guide; never clamp caller dimensions.
Validation: default/exact-boundary schema cases, prelaunch MCP rejection, compiled CLI capture, terminal rendering, normalization, cancellation, and lifecycle checks pass on macOS. This is resource admission, not a measured serializer throughput improvement; Linux-only and real deep-analysis providers were not exercised.
BREAKING CHANGE: process scenarios whose initial or resized terminal exceeds 1,000,000 combined visible-plus-scrollback cells are now rejected before launch.
Track created directory and file identities before materializing or removing output. Refuse displaced paths and untracked entries, preserve completed cleanup work across permission failures, and prevent publication once rollback starts. Group cleanup files once per attempt instead of rescanning them for every directory.
Validation: root, nested-parent, and file symlink replacement; untracked-entry retention; real POSIX permission failure and retry; extraction integrity and CLI/MCP consumers pass. Path identity revalidation cannot eliminate syscall-boundary pathname races in portable Node. Windows behavior was simulated; native Windows and Linux-only publication providers were not exercised.
Accumulate observations once per structural path instead of allocating and serializing a path for every payload value. Remove the CDP body-shape wrapper and retain primitive, empty-array, and literal-property semantics.
Validation: CDP capture/comparison regressions and CLI/MCP boundary checks pass; 1,507 differential cases match the previous implementation. A 2.5 MB repeated-row workload improves from 105.0 ms to 14.6 ms warm median across alternating runs. Real browser providers were not exercised.
Guide task-centered recovery from observed client, registration, PATH and
provider blockers. Install the recovery reference, document exact JavaScript
module selectors, and require clarification before searching for an unnamed
app. Validate public setup in isolated Codex accounts and preserve failed
or unrecognized tool outcomes in the existing evaluator.
Refs: #1587
BREAKING CHANGE: Optional verify:agent reports now use schemaVersion 4.
Consumers must accept expectedFirstTool: null and targetClarificationPassed
for missing-target scenarios. Production CLI/MCP contracts are unchanged.
Interrupt optional CDP domain shutdown on caller cancellation while preserving an independent bounded detach attempt. Retain original overflow and cleanup failures, verify workspace recovery, preserve user skill files, and reject another unverified firmware worker.
Validation: 149 integrated tests; repeated 64 affected boundary tests; full local check. Standalone CI includes real Chrome CLI/stdio MCP and real Web runtime checks.
Refs #1613
Validate declared artifact and MVID scopes at managed Evidence admission, and reject conflicting known module declarations across graph inputs. Join boundary tokens to members only under matching known MVID scopes; preserve facts and report partial coverage when scope is unknown.
Prefer an observed MVID-bearing declaration with its supplying Evidence. Normalize GUID casing for module node identity while preserving original strings in observations.
Validation: 285 tests across 34 focused suites passed, followed by the final 12-test managed graph suite and managed CLI/MCP regressions. check:fast and architecture guards passed; no real native providers were run.
Give Ghidra and Hopper an 8 MiB aggregate retained-output budget while continuing to drain their process pipes. Expose total observed bytes at the supervisor boundary and retained bytes plus explicit truncation in provider diagnostics.
Redact known transport tokens even when the retained prefix ends inside one, preserving adjacent caller path/query evidence. Existing partial-launch and cleanup ownership remain intact.
Validation: 285 tests across 34 focused suites passed; the final cutoff-fixture refinement passed all 31 Ghidra/Hopper diagnostic tests. check:fast and architecture guards passed. Provider fixtures use real child processes; real Ghidra/Hopper applications were not verified.
Use the header size already validated against the owned archive snapshot instead of asking the ASAR SDK to read and allocate its raw header again.
Validation: ASAR freshness, containment, streaming-allocation, and MCP extraction suites passed within the 285-test focused run. check:fast and architecture guards passed.
Pass validated coverage into normalization and index retained sources by script ID instead of scanning them for every script. Preserve source revalidation, UTF-16 bounds, nested counts, and unavailable-source uncertainty.
Validation: 285 tests across 34 focused suites, check:fast, and architecture guards passed. Flameox-guided 8,000-script normalization benchmark improved from 101.6 ms to 5.5 ms median with identical full-result digests; this is normalization CPU, not end-to-end browser timing.
Scan eligible candidate relations directly for query ambiguity instead of allocating and sorting another traversal index. Preserve direction, relation filters and cycle handling.
Semantic graph and builder suites plus public application workflow CLI tests passed. Full query outputs matched on a generated 20000-node graph; the single repeated-query timing run is indicative only.
Walk delimiter-terminated parent segments lazily instead of splitting full paths before discovering that no deeper observed parent exists. Preserve canonical paths and observed ancestry.
Verified real ZIP inventory with multiple 30000-component siblings and the full entry-order suite. A 96-member deep ZIP benchmark retained an identical full inventory digest and reduced median scan time from 517.5 to 464.5 ms across five measured inventories.
Resolve current paths and JVM options and rehash source identities for each request. Probe Java modules and engine capabilities only when creating a new session; remove the superseded full configuration resolver.
Focused configuration, session retirement, cancellation, cleanup and Android MCP boundaries passed. The 34-operation retained session now performs one prerequisite probe. Protocol fixtures do not establish real JADX or JDK performance.
Recheck unreadable token and identity observations against one fresh live process snapshot per phase. Preserve token mismatches, fail closed on inspection errors, and retain liveness failure diagnostics.
Validated with focused process ownership, native capture, CLI and MCP boundaries. An eight-member injected-unreadable workload using native ps reduced snapshots from nine to two with equivalent failures; real permission-denied provider behavior was not exercised.
morluto