OpenCode V2 reads native server entries from mcp.servers, and a native
entry takes precedence over a V1 mcp.<name> entry. After OpenCode
migrated a configuration, setup wrote an ignored V1 entry, status
reported REA as missing, and uninstall left the native entry in place.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(ghidra): admit the 12.1 line and the install's JDK range
The provider treated the verified Ghidra 12.1.4 build and JDK 21 as the only
legal installation, so a working 12.1.x package or a newer JDK was rejected
before analysis started.
Admission now follows the 12.1 release line and application.java.min/max.
A compatible build other than 12.1.4 keeps its observed version and records
that the bridge was verified against 12.1.4.
Fixes#821
* fix(ghidra): stop pinning the NativeAOT build to one version string
The build script still rejected every Ghidra release except the literal
12.1.4 and every javac except major 21. It now accepts the 12.1 line and
javac 21 or newer, compiles with --release 21, and records the observed
Ghidra version. The JADX engine version uses the audited release constant.
* fix(ghidra): report unverified builds on analysis results
Session open uses health only as a startup gate and discards that
execution, so the unverified-build limitation never reached CLI or MCP
results. Attach it to every Ghidra execution for that install.
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
* fix(firmware): accept the verified Binwalk and Unblob release lines
Exact --version equality rejected every build except Binwalk 3.1.0 and
Unblob 26.6.4, then stored that constant as the observed engine version.
Admit the 3.1 and 26.6 lines, keep the audited build as verified, and
report the banner that actually ran.
* chore(docs): regenerate skill metadata after firmware contract text
docs:check failed because SKILL.md lagged the Binwalk/Unblob release-line
tool descriptions.
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Artifact path failures told callers to "inspect the reported path" but
reported none: `ArtifactReaderFailure` messages such as
`Artifact path collision: ...` were dropped when converted to
`ArtifactOperationError`, leaving only `{ operation, reason: "path" }`.
Carry the failure message into `details.detail` for artifact tools and
JavaScript application analysis. Name both spellings in case-fold
collisions and include the JSON-escaped input in unsafe and
non-normalized path messages.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The XML fallback inferred an integer from `Number.isInteger`, so an
integral `<real>` beyond 2^53 was misreported as an unknown integer.
Read the `<integer>` and `<real>` literals from plutil's XML text instead:
an unsafe integral number that matches one `<integer>` literal and no
`<real>` gets its exact decimal; a number that could be either stays as
decoded with a limitation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`inspect_plist` parsed plutil JSON with `JSON.parse`, silently rounding
integers outside ±(2^53 - 1): `18446744073709551615` became
`18446744073709552000` with no limitation. Keep the exact decimal text
from the JSON source as `{ "$plist_type": "integer", "decimal": "..." }`
and add a limitation. On the XML fallback path, the decoder has already
rounded such values, so they become `decimal: null` instead of a wrong
number.
Also restore the `` escape in the BOM regex, which an earlier
rewrite had turned into a raw invisible character.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Explain that bare `rea doctor` audits every integration, map common tasks
to `--provider`, `--client`, and `--skill` scopes, and describe how
`scope_checks`, `informational_checks`, `healthy`, and
`environment_healthy` relate. Document the typed ambiguous-provider
error and how to choose an engine once without implying a fallback.
Fixes#726
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`inspect_plist` converted every plist with `plutil -convert json`, which
rejects plists containing `<data>`, `<date>`, or non-finite `<real>`
values. Such valid plists (including Stocks.app's Info.plist) failed with
a cause-free `execution_failure`.
When the JSON conversion exits nonzero, decode `plutil -convert xml1`
instead. Data and dates use the `$plist_type` objects the keyed-archive
reader already emits; non-finite reals become an explicit unknown because
the XML decoder reports NaN and infinity alike. Provenance records both
conversions and limitations explain the representation. Plists that JSON
can express keep the unchanged single-conversion path.
The projection moves to `src/domain/plistValue.ts` and is shared with the
keyed-archive reader, which keeps rejecting non-finite reals.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A source map whose `sources` contains `""` (a legal relative reference,
emitted for virtual or generated modules) made the whole application
analysis fail: the empty name became the `source-map-original` identity's
`original_source`, which must be nonempty. Display it as `""` at the same
accumulator seam; the observation's `properties.source` keeps the exact
empty name.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A single legal empty literal such as `fetch('')`, `require('')`,
`bus.on('', f)`, `localStorage.getItem('')`, or `ipcMain.handle('', f)`
made the whole application analysis fail with `unreadable_output`: the
literal became an application-graph observation label, an artifact-local
identity key, or a semantic node label, all of which must be nonempty.
Normalize display at the two graph construction seams instead of at each
producer: `JavaScriptArtifactGraphAccumulator.addNode` and
`constructSemanticGraphNode` show the empty value as `""` via a shared
`javascriptDisplayText` helper. Observation and node properties keep the
exact empty value. This replaces the per-site `semanticNameLabel` calls.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Document the three disjoint failure taxonomies (filesystem reader
codes, session reference reasons, inventory TypeErrors) with a test
pinning their disjointness. Export ParsedInventorySet and add
tryAssembleInventorySet covering the empty-pages leaf as Result;
the throwing entry contract is unchanged until callers migrate.
Delete the private canonicalJson duplicate in
functionComparisonNormalization in favor of comparisonSemantics
(same canonicalize behavior via the context parameter).
Add comparisonStatus as the single owner of the canonical
unchanged/changed/unknown triple and conclusion vocabulary, with
mappers pinning every pipeline-native status; adopt the shared
triple schema where wire-identical (bundle result, scenario
overall/step). Native detail enums and collation choices stay
untouched.
Migrate the remaining `""`-sentinel consumers from #783 to the nullable
`semanticStaticPropertyKey`: CommonJS module origins
(`require("m")['']`), require destructuring (`const {'': a} =
require("m")`), and request body fields (`{ '': value }`). Dynamic keys
still produce no link or field. API selectors that compare against known
names are unchanged.
Fixes#783
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Record the pinned Node 24.18.0 setup and that Hopper, Ghidra, and IDA stay optional for JavaScript analysis and deterministic tests.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
`process.env['']` now projects an exact empty environment key with
complete resolution instead of the same null/partial result as a dynamic
`process.env[key]`. `{ ['']() {} }` keeps its exact empty method name in
the IR instead of a generated `[computed@N]` name.
Semantic graph labels stay nonempty: a shared `semanticNameLabel` helper
displays the empty key as `""` for property slots, config sources, and
function nodes, while `properties.key` keeps the exact value.
Refs #783
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Report canonical BrowserExclusionReasons from runtime location
authorization (unsupported_url, not_approved) and collapse them
onto the fixed V8 wire bucket via inspectorExclusionKey, keeping
describeInspectorTargetLimitations text stable. Document the
provider authority split (attach-only CDP/V8 vs owned-runtime
Playwright) so the shared vocabulary is not re-split.
Move hasScheme/stripQueryAndFragment/admitsCanonicalPathSyntax
plus looksExternal into domain/artifactPathSyntax with the
deliberate policy-divergence table (artifact root vs allowedOrigins
vs exact-origin fetch). Adopt in JavaScriptArtifactPathResolution
and the Electron native specifier check with characterization
tests. Policy unification tracked in #829.
Move the indexed sections walk shared by the browser fetcher and
the artifact reader into domain/sourceMapEnvelope with the
fetchers stricter offset validation as an option; both validators
keep their leaf strictness. Rename the browser CDP stringValue to
cdpStringValue across its 18 files, fold the fingerprint duplicate
into the static helper, and rename the version-keys shadow to
nonEmptyString.
Inline the private memberKey/propertyKey wrappers in semantic
MemberMutations/Projection/Values to direct canonical calls.
Migrate the electron Browser/Values member keys off legacy
propertyName (provably identical: explicit computed guard retained).
Delegate bundle calleeName/endpointArgument/stringArgument to the
static helpers, adding loadURL endpoint parity with static analysis.
Bundle regressions pin nested-dynamic-receiver and loadURL parity.
Remaining computed-identifier forks in electron Native/Ipc tracked
in #822.
Iwakurakaoru0822
N0zoM1z0
morluto
Rudy Mizrahi Celekli