Keep caller-selected replacement text literal with a callback, preserving
empty-search positions instead of changing their split/join behavior.
Commit the replacement interpretation for captures using custom patterns so
legacy records cannot report this producer correction as a target change.
BREAKING CHANGE: Captures with custom normalization patterns use
literal-replacements-v1 and cannot be compared with legacy unversioned
pattern captures. Capture both sides with the same replacement semantics;
legacy captures remain readable and captures without patterns are unaffected.
Strict JSON file inputs reject a leading UTF-8 byte-order mark, but the
diagnostic was the parser's own message, which quotes the invisible U+FEFF
as the unexpected token: Unexpected token '', "{"invento"... is not
valid JSON. Editors such as Windows PowerShell 5.1 write that mark by
default, and nothing in the message showed it.
CLI JSON files (whole and streamed), Evidence and snapshot JSON files, and
process scenario files now report that the input begins with a UTF-8
byte-order mark and how to save it. Acceptance is unchanged.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Keep the Bahasa Indonesia onboarding text from #1508, and fix the formatting
that left it outside the shared README layout: drop the leading space, point
in-page links at the translated headings, align the analysis table, and link
Bahasa Indonesia from every language selector.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Dev Cee <24010110166@universitasbumigora.ac.id>
* feat(setup): add native Pi client support
* fix(setup): isolate invalid Pi agent directory overrides
Keep malformed Pi path evidence on its client entry so unrelated selected clients can complete setup and doctor. Preserve explicit Pi rejection and block filesystem operations rather than falling back to another directory.
* fix(setup): parse Pi configuration as strict JSON
Pi's installed MCP loader reads mcp.json with a bare JSON.parse and no
BOM tolerance, so comments, trailing commas, a BOM, or an empty existing
file yield loader errors and no servers. Parsing the Pi dialect with
objectSchema.parse(JSON.parse(text)) before the shared empty/JSONC
shortcut rejects those files at every mutation boundary instead of
reporting configured registrations Pi cannot load. A missing file still
initializes a fresh strict-JSON document, other dialects keep their
JSONC/BOM/empty tolerance, and malformed files are never rewritten.
---------
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
* feat: add Hermes as a supported MCP client
Hermes Agent reads stdio MCP servers from `mcp_servers.<name>` in
`config.yaml`. REA had no Hermes client, so setup could not register it and
Hermes users had to wire the server up by hand.
- SupportedClients: add the `hermes` format and client definition, resolving
the config path from HERMES_HOME, else the platform default that Hermes's own
resolver uses (LOCALAPPDATA/hermes on Windows, ~/.hermes elsewhere), including
HERMES_DATA_DIR_SUFFIX.
- ClientConfigurationDocument: parse and serialize Hermes YAML, upserting the
rea entry into the parsed document so user comments and unrelated settings are
preserved; emit `enabled: true` since Hermes connects entries unless disabled.
- ClientRegistrationStatus: forward the Hermes location variables so status
checks resolve the same path as setup.
- Tests: cover path resolution on both platforms and the register/read/uninstall
round trip, and clear the Hermes location variables between tests.
- delilah 🌸
* docs: document the Hermes client and its config path
- delilah 🌸
* refactor(setup): resolve skill destinations per client
* fix(setup): install Hermes skills in the active profile
* fix(setup): identify each client skill action uniquely
---------
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
Every inventory edge is created from one occurrence and copied that
occurrence's artifact_id and logical_path as child_artifact_id and
logical_path. Edges now keep the occurrence_id reference alone. Edge IDs
still hash the same child and path, read from the occurrence, so they are
unchanged; validation and artifact comparison read the occurrence too.
inspect_artifact output: Transmission.app 1.10 MB to 0.97 MB, Visual
Studio Code.app 15.37 MB to 13.48 MB.
BREAKING CHANGE: Artifact inventory edges no longer include
child_artifact_id or logical_path; read them from the occurrence named by
occurrence_id. The graph commitment changes, so inventory Evidence
recorded by earlier versions no longer validates.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
inventory_artifact, inspect_artifact, and extract_artifact copied each
occurrence or extracted file path into Evidence locations, although the
result already names every one. inspect_artifact did so twice, in its own
Evidence and in the nested inventory Evidence. Like other whole-target
Evidence, these now report no locations.
Visual Studio Code.app: inspect_artifact 15.37 MB to 13.07 MB,
extract_artifact 1.40 MB to 1.06 MB.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Input schema presentation gave every undescribed property a generic
"Value for <name>." sentence. That only restates the property name, yet
it covered 1,958 properties: 70 KB of the 522 KB of advertised input
schemas that clients load into model context. Authored descriptions,
referenced descriptions, and the name-pattern sentences that add meaning
(maximums, digests, paths, Evidence IDs) remain; other properties now
advertise no description.
The schema validity test now rejects placeholder descriptions instead of
requiring a description on every property.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
ZIP member names are case-sensitive. Inventory was folding them with a
filesystem case rule, so shrunk APKs never became evidence for the Android
application graph. Logical registration now keeps both spellings and records
an English Unicode case-folding limitation. Extraction fails only when the
destination directory already holds the other spelling.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
inspect_artifact nested the complete inventory Evidence and then restated
every artifact node and occurrence as an observation and every edge as a
derived relationship. No caller reads those copies; comparison and tests
use the nested inventory Evidence. On Transmission.app the result shrinks
from 2.46 MB to 1.10 MB.
Observations keep the root manifest and integrity findings.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Include the full stop disposition in the fixture assertion message, so incomplete owned teardown retains its typed reason in Vitest failure reporters. Preserve the existing cleanup requirement without retries or relaxed ownership checks.
Seven process-boundary tests, full serial typecheck/lint/duplication/format checks, and 19 final-head CI checks pass after integrating the combined-main fixture fix. A controlled failure-rendering probe verifies that the explicit message preserves the reason. The original CI cleanup failure remains unreproduced; this is diagnostic work with a synthetic JADX producer, not a root-cause or real-engine fix.
Refs #1441, #1446, #1448, #1065.
Validate the target-free probe against the named definition discovered from the complete production catalog. Preserve initialized server and inventory observations when the SDK rejects the probe, and retain the schema failure reason, absolute deadline and owned-child cleanup.
The negative stdio replay regression fails on original main and passes the fix; actual production-boundary fault injection confirms rejection. Nine doctor/help tests, serial full static checks and real small-target CLI/MCP parity pass. Successful doctor reports remain byte-identical. No performance or model-token reduction is claimed.
Closes#1454. Refs #1065, #723, #1433.
Rendered frames padded every visible row with trailing spaces to the
terminal width. Blank rows cost a full row each and every frame repeats
them, so the padding made up 14-39% of capture Evidence (2 KB of sparse
output produced 228 KB). The padding carried no information: columns is
recorded and serialized_state keeps the exact terminal state.
Lines are now trimmed after normalization, so line.padEnd(columns)
restores the previous value. The comparison contract commits
rendered_line_format so older padded captures do not compare as the
same contract.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Load browser composition after CLI scenario validation and load Git support only for local reference trees with Git metadata. Preserve default outputs and actual command behavior.
Refs #1065.
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Advertise content- and dialect-bound identities for self-contained output schemas so the SDK can reuse unchanged Ajv validators after complete tools/list refreshes. Preserve explicit IDs and relative reference bases.
Verify changed schemas, availability transitions, reconnects, numeric and selector constraints, explicit IDs, relative references, and advertised dialects through the pinned SDK boundaries.
Refs #1065, #723
inspect_macho, inspect_signature, inspect_plist, list_architectures and
demangle_swift already return their command provenance inside the
normalized result. The provider also copied the same array into
raw_result, so every response carried it twice. raw_result now stays
null, as docs/mcp-contracts.md describes for results without a distinct
provider representation.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Share the canonical string identity rule with endpoint projection, scope event keys to each emitter, and keep complete event/listener names in properties. Verify real CLI and stdio MCP with large names, registration/removal, distinct emitters, exact name seeds, reserved-prefix values, and unresolved dynamic names.
BREAKING CHANGE: New analyses use digest-based IDs for long or reserved-prefix event names and shorter event/listener display labels. Select IDs from the parent graph and read exact names from event_name properties. Stored graphs and their original IDs remain accepted for semantic tracing.
Keep exact endpoint observations and request properties while using compact display labels and collision-safe canonical value digests for long artifact-local endpoint keys. Verify real CLI/stdio MCP tracing and version comparison without truncating evidence.
BREAKING CHANGE: New analyses use compact endpoint labels and digest-based IDs for long or reserved-prefix endpoint keys. Select IDs from the parent graph and read full values from endpoint properties. Previously stored graphs retain their original identities and remain valid workflow inputs.
`--json` / `--format json` always indented its streamed result. Agents and
scripts read REA through pipes and redirects, where indentation carries no
information: on real application results it added 16-45% model tokens
(o200k proxy) and 25-73% bytes. The skill's CLI fallback uses `--json`.
Indent only when stdout is a TTY, as `gh --json` does; piped or redirected
JSON is compact. The JSON value, `jsonl`, `--filter-output`,
`--full-output` and the token-window paths are unchanged. `jq .` restores
indentation for files.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Commit literal values with canonical JSON SHA-256 in semantic role keys
and use a type label for strings. Preserve complete values, provenance and
literal queries while avoiding two payload copies in each string node.
Verify real compiled CLI/default stdio MCP, typed union seeds, empty
strings and trace continuation with a large source-owned fixture.
Refs #723
BREAKING CHANGE: Newly produced semantic literal node IDs use
value-sha256 role keys, and string labels read "string literal". Read exact
values from properties.value and obtain node IDs from the selected parent
graph. Previously stored Evidence and its original IDs remain valid.
HopperProvider passed the bridge reply as rawResult and, for every
operation except address_to_file_offset, the same object as the result.
Every Hopper Evidence record (decompilation, procedure and string
listings, xrefs, ...) therefore carried its complete result twice, and
an MCP response four times counting its text copy.
raw_result is now null unless REA transformed the reply, which only the
file-offset mapping does; that case keeps Hopper's original reply.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
inspect_binary_layout and inspect_recorded_crash passed the decoded report
as both result and rawResult, so every Evidence record carried the whole
report twice (and an MCP response four times, counting its text copy).
The decoded report is the only representation, so raw_result is now null,
as for other single-representation operations.
The real pwntools verifiers asserted the duplicate; they now assert null.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
createAnalysisExecution defaulted an omitted rawResult to a second
complete copy of the result. Every caller that had no distinct provider
representation therefore doubled its Evidence: decode_interface_builder,
inspect_keyed_archive, inspect_asset_catalog, trace_dylib_resolution,
inspect_native_dispatch_metadata, observe_native_calls,
observe_native_ui/capture_native_ui_scenario (including screenshots), and
small Ghidra/IDA lifecycle records. Sibling operations of the same
providers already recorded null.
An omitted rawResult now records null. Normalized results are unchanged.
MCP responses repeat Evidence in text and structured content, so each
response and its share of the 10 MiB transport budget shrink by the same
half; e.g. trace_dylib_resolution on a real app bundle drops from
3,217,119 to 1,612,524 bytes per representation.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
compare-web-captures and compare-web-screenshots parsed their arguments as
inline JSON only, unlike every other JSON-input command. Screenshot
artifacts and scenario captures embed base64 PNG bytes, so a typical
viewport screenshot exceeds the command-line length limit (128 KiB per
argument on Linux, 1 MiB total on macOS) and could not be compared from
the CLI at all, while MCP accepts it. A file path was reported as
malformed JSON, and screenshot schema failures returned no issues.
Each argument now accepts inline JSON or a JSON file path through
parseCliJsonInput; file failures keep its path diagnostics, malformed
inline JSON keeps the field issue, and screenshot validation reports
projected schema issues. The tests share one CLI JSON process fixture.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
trace_web_module_imports and trace_web_source_location reported a missing
or non-file manifest, source map or import map as artifact_operation_failed
("Artifact could not be read or written" / "unsafe or conflicting internal
path"). These are caller selections: report ENOENT/ENOTDIR and non-regular
files as invalid input on the selected field, and EACCES/EPERM as a host
access denial. A missing manifest-derived source file stays an artifact fault.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Incur's `alias` map registers one-letter short flags and already accepts
kebab-case spellings of camelCase options. Twenty options registered their
kebab-case long name as an "alias", so help and completions advertised
flags such as `-case-sensitive` and `-relative-path` that the parser reads
as stacked short flags and rejects with "Unknown flag: -c".
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Electron recognition matched callee spellings, so a bundler rename such as
esbuild's `import { ipcMain as ipcMain2 } from "electron"`, or a destructured
`const { contextBridge: bridge } = require("electron")`, hid every IPC
handler, BrowserWindow, contextBridge and utility process behind it.
Resolve each call and construction root lexically to its Electron import or
require origin, and classify the callee under that export's name. Shadowed
locals keep their own meaning.
On Hermes Desktop v0.21.6, whose main process is one esbuild ESM bundle,
recognized main IPC handlers rise from 1 to 204, BrowserWindow constructions
from 0 to 11, and paired renderer transmissions from 1 to 201.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XPY2GmufMNGQKoEgRUx1pd
The cross-format oracle requires a callee whose recovered control flow
has more than one basic block, and rea_cross_branch provided it through
a symmetric if/else of tail calls. Clang 21 folds that shape into a
conditional negate plus one tail branch, a single basic block, so the
lane rejected a correct Ghidra analysis of the fixture.
Give the branch a loop whose trip count depends on a volatile global:
no compiler can eliminate the backedge, and the callee stays multi-block
at every optimization level on every toolchain.
* fix(mcp): advertise object roots for union input schemas
Anthropic tool input schemas reject anyOf, oneOf, and allOf at the root.
Since 63bf9140 removed root flattening, compare_web_captures,
inspect_web_page, inspect_electron_page, and
project_managed_application_graph advertised root unions: Claude Desktop
disabled compare_web_captures and Claude Code exposed the tools without
parameters.
Present a root union of plain-reference object branches as one strict
object with the union of properties, shared required fields, a
minProperties bound, and a description of the accepted input groups. The
canonical Zod union still validates every call, so mixed input groups are
rejected before the handler runs. Branches whose reference siblings cannot
be merged keep their root alternatives.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(browser): keep comparison input boundary within lint limits
Express advertised and runtime expectations as one case table so the
boundary test stays under the 150-line function limit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(mcp): preserve input group dependencies in object schemas
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
* fix(javascript): preserve uncertainty after references escape to calls
* fix(javascript): track iterator root escapes and preserve static captures
* fix(javascript): scope comparison unknowns to their evidence
* fix(javascript): track parameter origins at function body entry
---------
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
Select residual unknown histories as complete units and retain referenced registry mutation Evidence regardless of the active target at edit time. Prune incompatible dependency histories atomically instead of retaining stale dispositions.
Verify public MCP save/read/reopen for verified, withdrawn, and out-of-scope resolutions, and document target scope rules.
* fix(javascript): decline source-map UNC references with an unusable host
A source map can name a Windows share whose host a file URL cannot carry,
such as `\\bad host\share\a.js`. On the Node builds `engines` admits,
`pathToFileURL(path, { windows: true })` fails a native assertion and calls
`abort()` on Node 22.19-22.22, 24.11-24.15 and 26.0, so the process dies
with exit 134 instead of declining the reference. Node 22.23 and 24.16 raise
`ERR_INVALID_URL` instead, and older builds silently drop the host, so the
same reference resolves three different ways depending on the runtime.
Validate the UNC host before handing the path to Node. A host carrying a
forbidden host code point, a percent escape or nothing at all now resolves as
`unresolved` on every admitted build, and `npm test` passes the existing
`\\bad host\share\a.js` row everywhere. Well-formed UNC hosts and
drive-letter paths are untouched.
* fix(javascript): validate UNC hosts through IDNA before conversion
Reject Unicode source-map UNC hosts that URL/IDNA processing cannot use or normalizes to an empty hostname before calling Windows pathToFileURL. Keep the original source identity and preserve usable normalized hosts.
Exercise the built reference resolver in owned subprocesses, including U+00AD and U+200C as source names and source roots. The original PR implementation aborts under Node 22.22.2; the new regression isolates those signals from the test runner. Add this test to the existing Node 22 Windows CI step.
Validation: 64 focused tests pass on Node 22.22.2 and 24.18.0; npm run check and test discovery pass. Local verification ran on macOS; Windows verification remains with CI.
---------
Co-authored-by: Yi-111-a <yi-111-a@users.noreply.github.com>
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
N0zoM1z0