Critical-review §4 E15 finding: agent files in .claude/agents/ are loaded
as Claude Code subagent system prompts and are a direct memory-poisoning
surface. Pre-v7.2.0 the scanner covered CLAUDE.md, .claude/rules/*.md,
memory/*.md, REMEMBER.md, .local.md, and .claude-plugin/plugin.json —
but not .claude/agents/*.md.
Single-line addition to MEMORY_FILE_PATTERNS:
/(?:^|\/)\.claude\/agents\/[^/]+\.md$/
The existing scan loop, scanForInjection integration, and severity-
mapping logic all apply unchanged. STRICT_FILES_PATTERN intentionally
NOT extended — agents may legitimately quote shell commands as examples
(consistent with CLAUDE.md treatment).
Tests: +3 cases in tests/scanners/memory-poisoning.test.mjs:
- "scans .claude/agents/*.md" (smoke test — at least one finding from
the new fixture)
- "agent file injection pattern detected"
- "agent file credential path detected"
New fixture: tests/fixtures/memory-scan/poisoned-project/.claude/agents/
poisoned-agent.md — agent with injection, credential ref, permission
expansion, and exfil URL. Triggers all 4 detection categories.
Suite: 1591 → 1594 (+3). All green.
Critical-review §2 B7 finding: pure Levenshtein <=2 misses the most common
modern typosquat pattern — popular-name + token-injection suffix. Examples:
lodash → lodash-utils (edit distance 6, not flagged pre-B7)
react → react-helper (edit distance 7, not flagged pre-B7)
express → express-wrapper (edit distance 8, not flagged pre-B7)
Three coordinated edits:
scanners/lib/string-utils.mjs
- Adds tokenize(name): string[] splits on -/_, lowercases
- Adds tokenOverlap(a, b): number intersection.size / min(|a|,|b|)
- Adds TYPOSQUAT_SUSPICIOUS_TOKENS frozen list of common typosquat
suffixes. Excludes language-extension tokens (js, jsx, ts, tsx) — the
v7.0.0 allowlist contains `tsx` as a legit package and including the
same token in the suspicious set creates a contradiction. Caught by
the new allowlist-intersection-guard test. Also excludes 'pro'
(legitimate edition marker).
scanners/dep-auditor.mjs + scanners/supply-chain-recheck.mjs
- New checkTyposquatTokenOverlap() helper — fires AFTER Levenshtein 1/2
branches, only when:
1. popular package's tokens ⊆ declared name's tokens (strict superset)
2. declared name has at least one suspicious suffix
3. popular package is in topCutoff window
All three conditions required — conservative by design. Allowlist
precedence preserved (existing 22 npm + 13 PyPI entries always pass).
MEDIUM severity, NOT block. New finding title prefix:
"Possible typosquatting via token-overlap".
Tests: +21 cases across two new files
- tests/lib/string-utils-tokens.test.mjs (15) — tokenize, tokenOverlap,
TYPOSQUAT_SUSPICIOUS_TOKENS frozen contract, allowlist-intersection
guard (caught the tsx conflict on first run)
- tests/scanners/dep-token-overlap.test.mjs (7) — integration via
in-memory tmpdir fixtures: lodash-utils flagged, react-helper flagged,
express-wrapper flagged, lodash exact NOT flagged, allowlist tools
(knip/tsx/nx/rimraf) NOT flagged, react-router-dom (no suspicious
suffix) NOT flagged, react itself (equal token set, not superset)
NOT flagged.
Existing dep.test.mjs and supply-chain-recheck.test.mjs unchanged —
all green (149 → 149 regression guard).
Suite: 1570 → 1591 (+21). All green.
Critical-review §2 B6 finding: extractAssignedVariable handled
`const X = ...` and `X = ...` but missed every modern JS/TS
destructuring pattern. Sinks downstream of destructured/spread vars
produced false negatives at the propagation step.
Patterns now recognized:
- `const { x } = source` object destructuring
- `const { x, y } = source` multi-key
- `const { secret: alias } = source` renamed (key NOT bound)
- `const { x, ...spread } = source` object rest
- `const { a, b: { c } } = source` nested object (key NOT bound)
- `const [a, b] = source` array destructuring
- `const [first, ...rest] = source` array rest
- `const [a, [b, c]] = source` nested array
- `const { user: { id }, ...rest }` mixed nested
Implementation: regex-based two-pass walker. Pass 1 detects whether
the LHS is a destructuring pattern (`{...}` or `[...]`). If yes, the
new `extractDestructuredNames` helper walks the pattern body via a
balanced-bracket depth counter, recurses into nested patterns, and
distinguishes keys (`key:`) from bindings. If no, the plain-decl
branch matches `\b(?:const|let|var)\s+(\w+)`.
Plain-assignment branch (`X = ...` without keyword) and Python-style
patterns are unchanged.
The function is now exported for direct unit testing — same pattern
as `_resetCacheForTest` in policy-loader. The internal walker
(`extractDestructuredNames`) remains module-private.
Tests: +19 cases in tests/scanners/taint-destructuring.test.mjs:
- 5 pre-B6 patterns (regression guard: plain decl, plain assign,
no-match on equality)
- 12 destructuring patterns covering object/array/rest/nested
- 2 non-destructuring regressions (return literal, arrow param)
Existing taint-tracer.test.mjs and taint.test.mjs unchanged — both
green (14 → 14, fixture-based integration tests not affected).
Suite: 1551 → 1570 (+19). All green.
Critical-review §2 B3 finding: `riskScore({info: N}) = 0` silently masks
info-volume findings. The behavior was correct (info is scoring-inert by
design) but undocumented. Operators reading a report with N info findings
had no way to know they contribute zero to verdict/band.
Three coordinated edits:
- scanners/lib/severity.mjs JSDoc — explicit "Info severity" subsection
spelling out: scoring-inert, surfaced in owaspCategorize aggregates,
treat as observability telemetry not verdict input. @param updated to
mark info as accepted but ignored.
- CLAUDE.md v7.0.0 risk-score-v2 line — one-sentence anchor pointing to
severity.mjs JSDoc.
- tests/lib/severity.test.mjs — anchor test alongside the existing
4-critical=93 anchor: asserts riskScore({info: 50}) === 0,
riskScore({info: 1000}) === 0, verdict({info: 100}) === 'ALLOW',
riskBand(riskScore({info: 500})) === 'Low'.
Decision: skip the optional `infoScore()` helper from the brief. No
current consumer would use it; doc-only fix keeps API surface minimal.
Revisit if a consumer emerges.
Tests: 1522 → 1523 (+1 anchor block, 4 assertions). All green.
Closes A2 of v7.1.0 critical-review patch (docs/critical-review-2026-04-20.md):
- B4 (severity JSDoc): 4 critical = 93, not 90. Fixed in scanners/lib/severity.mjs:23
and CHANGELOG.md v7.0.0 tier description. The actual computation has always been
93 (70 + log2(5)*10 = 93.22 → round); only the docs were wrong.
- §5.4 co-monotonicity: new sweep test in tests/lib/severity.test.mjs over 15
representative count vectors. Asserts that (verdict, riskBand) agree under the
v7.0.0 contract for every case — catches future drift between riskScore tiers,
verdict cutoffs, and riskBand cutoffs. Includes a B4 anchor test (riskScore
{critical: 4} === 93) so doc/code drift fails loudly.
- B8 (CaMeL claims toned down): post-session-guard.mjs:646 comment block and
CLAUDE.md:184 Defense Philosophy bullet now describe the implementation
honestly — opportunistic byte-matching of truncated output fingerprints
(first 200 bytes, SHA-256/16-hex), not semantic data-flow tracking.
Trivially bypassed by mutation, summarisation, or re-encoding. Inspired by
CaMeL (DeepMind 2025), but not a CaMeL capability-tracking implementation.
Tests: 1495 → 1511 (+16: 15 sweep cases + 1 B4 anchor). All green.
E2E verification against content-heavy repo (`content-claude-code`) revealed
413 entropy findings (8 HIGH / 405 MEDIUM) from markdown image CDN URLs in
JSON content indexes — e.g., ``.
These are legitimate content-repo artifacts, not credentials. The 40-char
hash segment in the CDN URL trips Shannon entropy (H=5.29 over 300 chars),
and rule 13 (inline <svg>) doesn't match since there's no literal `<svg>`
tag — the `.svg` is just a URL path suffix.
Added rule 18 `MARKDOWN_IMAGE = /!\[[^\]]*\]\(\s*https?:\/\//` — matches
`` / ``. Line-level (not string-level) so URL
is not over-specific.
E2E impact on `content-claude-code`:
- Before: BLOCK / 65 / 8H 437M 0L
- After: WARNING / 56 / 3H 427M 0L
Hyperframes unchanged: BLOCK / 80 / 1C 4H 92M — real CRITICAL SQL-injection
and HIGH findings still detected.
Tests: 2 new (positive + negative fixture) bringing entropy-context to 26,
total suite 1485 → 1487.
Docs updated to "rules 11-18" and "8 new line-suppression rules".
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Final commit in the trustworthy-scoring series. Bundles verdict cutoff
alignment, the last suite of tests, and all documentation touch-points
that quote version numbers or describe v7.0.0 behaviour.
Verdict/band co-monotonicity
- `scanners/lib/severity.mjs` — verdict cutoffs moved from 61/21 to 65/15
so `BLOCK >= 65`, `WARNING >= 15` locks onto the v2 riskBand() boundaries.
Prevents "BLOCK / Medium band" contradictions under the v2 formula.
Scanner hardening (bug fixes from v7.0.0 testing)
- `scanners/entropy-scanner.mjs` — `policy_source` now uses
`existsSync('.llm-security/policy.json')` instead of value-based check.
Old heuristic always reported 'policy.json' because DEFAULT_POLICY now
carries an `entropy.thresholds` section.
- `scanners/lib/file-discovery.mjs` — `.sass` and GPU shader extensions
(`.glsl, .frag, .vert, .shader, .wgsl`) added to TEXT_EXTENSIONS. Without
this, shader files were invisible to file-discovery, so they were never
counted as skipped by the entropy-scanner extension filter.
Tests
- `tests/scanners/entropy-context.test.mjs` (new, 24 tests) — A. File-ext
skip (4), B. Line-level rules 11-17 (8), C. Policy overrides (3).
Fixtures generate 80-char base64 payloads at runtime via
`crypto.randomBytes` to dodge the plugin's own pre-edit credential hook
on the test source.
- `tests/lib/severity.test.mjs` — rewritten with v2 scoring table (70
tests total, was 52).
- `tests/lib/output.test.mjs:243` — "1 critical = score 80" under v2
(was 25 under v1).
- Full suite: 1485/1485 green (was 1461).
Docs
- `CHANGELOG.md` — v7.0.0 entry with BREAKING CHANGES section.
- `README.md` (plugin + marketplace root) — version badge, history table,
plugin-card version string, test count.
- `CLAUDE.md` — header version, "v7.0.0 — Trustworthy scoring" summary
paragraph at the top.
- `docs/security-hardening-guide.md` — new section 6 "Calibration & false
positives" documenting v2 formula, context-aware entropy scanner,
typosquat allowlist, and §6.4 tuning workflow. Existing "Recommended
baseline" section renumbered to §7.
Version bump
- `6.6.0 -> 7.0.0` across package.json, .claude-plugin/plugin.json,
scanners/ide-extension-scanner.mjs VERSION const, README badge,
CLAUDE.md header, marketplace root README card.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds entropy section to DEFAULT_POLICY and wires it into entropy-scanner.
Users can now tune false-positive tradeoffs without forking the scanner.
Policy shape (.llm-security/policy.json):
entropy:
thresholds.{critical,high,medium}.{entropy,minLen} — numeric overrides
suppress_extensions[] — additive ext skip
suppress_line_patterns[] — additional regex
suppress_paths[] — relPath substrings
Wiring: entropy-scanner calls loadPolicy(targetPath) at scan entry (not
orchestrator-passed — avoids signature churn across 10 scanners). Module-
level state is reset per scan invocation. Scanner envelope now includes
calibration.{policy_source, thresholds, files_skipped_by_*} for
synthesizer transparency (Commit 5).
Malformed user regex silently skipped. Missing policy.json → built-in
defaults (backwards-compatible).
entropy.test.mjs: 9/9 still green.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Observed 70% false-positive rate on renderer/shader codebases (hyperframes):
GLSL, CSS-in-JS, inline HTML/SVG, ffmpeg filter-strings, hardcoded
User-Agent strings all matched base64-like entropy thresholds. This
commit adds two suppression layers before classification.
Layer A — file-extension skip: .glsl/.frag/.vert/.shader/.wgsl (shaders),
.css/.scss/.sass/.less (stylesheets), .svg (markup), .min.js/.min.css
(minified bundles). Tracked via new calibration.files_skipped_by_extension
field on scanner envelope for synthesizer stats.
Layer B — seven new line-level suppression rules in isFalsePositive()
(rules 11-17): GLSL/WGSL keywords, CSS-in-JS (styled/emotion/@keyframes),
inline HTML/SVG markup, ffmpeg filter-graph syntax, browser User-Agent,
SQL DDL/DML, error-message templates with embedded HTML.
Existing entropy.test.mjs: 9/9 still green — known bad base64 payload in
telemetry.mjs fixture still detected. Policy-driven thresholds wired in
Commit 3.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Replace sum-and-cap formula (every non-trivial scan → 100/Extreme) with
severity-dominated, log-scaled-within-tier model. Discriminates actual
risk: 1 critical = 80, 2 critical = 86, 17 high = 65. Hyperframes-class
rendering codebases no longer collapse to Extreme just from shader noise.
Changes:
- scanners/lib/severity.mjs: new riskScore() v2; keep riskScoreV1() for
reference; riskBand() cutoffs aligned (14/39/64/84).
- scanners/posture-scanner.mjs: delete inline duplicate formula, import
riskScore/riskBand/verdict from severity.mjs. Single source of truth.
Breaking: aggregate.risk_score semantics change. Batched with entropy
suppression (Commit 2+) under v7.0.0 bump in Commit 6. Do not release
individually — JSON consumers depend on scoring band stability.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Symmetric with the existing VS Code branch — the env var was only wired
into getVSCodeExtensionRoots(), so the plan's master verification
(`LLM_SECURITY_IDE_ROOTS=... --intellij-only`) reported 0 discovered
plugins. Adding the same fallback to discoverJetBrainsExtensions makes
both families honor the CLI override and closes the gap.
Step 1/17 of ultraplan-2026-04-17-jetbrains-ide-scan.
- Populate top-jetbrains-plugins.json with 56 canonical xmlIds (bundled +
popular third-party): com.intellij.java, org.jetbrains.kotlin,
com.jetbrains.python.community, org.rust.lang, com.github.copilot,
mobi.hsz.idea.gitignore, the legitimate-typo 'Lombook Plugin', etc.
- Add loadJetBrainsBlocklist() export mirroring loadVSCodeBlocklist shape.
Blocklist is empty by design — no public confirmed-malicious JetBrains
Marketplace plugins as of 2026-04-17.
- Add tests/scanners/ide-extension-data.test.mjs (9 tests, all pass).
- Fix cache bug in loadTopJetBrains: map normalizeId on cache-hit path too
(was previously unnormalized on second call).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
VSIX fetch + extract for URL targets now runs in a sub-process wrapped by
sandbox-exec (macOS) or bwrap (Linux), reusing the same primitives proven
by the v5.1 git-clone sandbox. Defense-in-depth — even if our own
zip-extract.mjs ever has a bypass, the kernel refuses any write outside
the per-scan temp directory.
New files:
- scanners/lib/vsix-fetch-worker.mjs — sub-process worker. Argv: --url
--tmpdir; emits one JSON line on stdout (ok/sha256/size/source/extRoot
or ok:false/error/code). Silent on stderr. Exit 0/1.
- scanners/lib/vsix-sandbox.mjs — wrapper. Exports buildSandboxProfile,
buildBwrapArgs, buildSandboxedWorker, runVsixWorker. 35s timeout, 1 MB
stdout cap.
Changes:
- scanners/ide-extension-scanner.mjs: fetchAndExtractVsixUrl is now
sandbox-aware (useSandbox option, default true). In-process logic
preserved as fallback. New meta.source.sandbox field:
'sandbox-exec' | 'bwrap' | 'none' | 'in-process'.
- scan(target, { useSandbox }) defaults to true; tests pass false because
globalThis.fetch mocks do not cross process boundaries.
- Windows fallback: in-process with meta.warnings advisory.
Tests:
- 8 new tests in tests/scanners/vsix-sandbox.test.mjs (per-platform
profile generation, worker arg construction, live worker exit
behavior on invalid URLs — no network).
- Existing URL tests updated to opt out of sandbox (useSandbox: false).
- 1344 → 1352 tests, all green.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Pre-installation verification of VS Code extensions via URL — fetch a remote
VSIX, extract it in a hardened sandbox, and run the existing IDE scanner
pipeline against it. No npm dependencies.
Sources:
- VS Code Marketplace (publisher.gallery.vsassets.io direct download)
- OpenVSX (open-vsx.org official API)
- Direct .vsix HTTPS URLs
Defenses:
- HTTPS-only, TLS verified, manual redirect with per-source host whitelist
- 30s total timeout via AbortController
- 50MB compressed cap, 500MB uncompressed, 100x expansion ratio
- Zero-dep ZIP extractor: zip-slip, absolute paths, drive letters, NUL bytes,
symlinks (Unix mode 0xA000), depth limits, ZIP64 rejected, encrypted rejected
- SHA-256 streamed during fetch, surfaced in meta.source
- Temp dir cleanup in all paths (try/finally)
Files:
- scanners/lib/vsix-fetch.mjs (HTTPS fetcher, host whitelist, streaming SHA-256)
- scanners/lib/zip-extract.mjs (zero-dep parser with hardening caps)
- knowledge/marketplace-api-notes.md (endpoint reference)
- 3 test files (48 tests added: vsix-fetch, zip-extract, ide-extension-url)
Tests: 1296 → 1344 (all green).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Masks non-empty '...' content before T5/T2-T4 run so literal strings such
as `echo '${IFS}'` are not rewritten. Empty '' pairs are stripped first
so c''u''rl -> curl evasion keeps resolving. ANSI-C $'...' is decoded
before masking.
Caught by the false-positive probe added in Step 3 of ultraplan-v6.2.0.
Add threshold-based exit codes (--fail-on <severity>) and compact
output mode (--compact) to scan-orchestrator and CLI. Pipeline
templates for GitHub Actions, Azure DevOps, GitLab CI with SARIF
upload. CI/CD guide with Schrems II/NSM compliance documentation.
npm publish preparation (files whitelist, .npmignore). Policy ci
section for distributable CI defaults. Version 6.1.0.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
process.exit() terminates before pipe buffers drain, truncating output
at 64KB when piped through another Node.js process on macOS. Affects
scan-orchestrator (SARIF output) and supply-chain-recheck-cli.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New policy-loader.mjs reads .llm-security/policy.json with deep-merge against
defaults that exactly match existing hardcoded values. Integrated into all 7 hooks:
- pre-prompt-inject-scan: injection.mode (env var still takes precedence)
- post-session-guard: trifecta.mode, window_size, long_horizon_window
- pre-edit-secrets: secrets.additional_patterns
- pre-bash-destructive: destructive.additional_blocked
- pre-write-pathguard: pathguard.additional_protected
- pre-install-supply-chain: supply_chain.additional_blocked_packages
- post-mcp-verify: mcp.volume_threshold_bytes, mcp.trusted_servers
Backward compatible: no policy file = identical behavior to v5.1.0.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New audit-trail.mjs writes structured events to LLM_SECURITY_AUDIT_LOG path.
Integrated into post-session-guard at 6 warning emission points: trifecta,
escalation-after-input, data flow, volume threshold, slow-burn, behavioral drift.
No-op when env var not set — zero overhead for existing users.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New sarif-formatter.mjs converts scan envelope to OASIS SARIF 2.1.0 standard.
Maps severity to SARIF levels, findings to results with locations and rules.
scan-orchestrator accepts --format sarif|json (default: json).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extends posture scanner from 13 to 16 categories with three governance/compliance
checks. New categories are advisory (not in CRITICAL_CATEGORIES) — existing Grade A
projects remain Grade A. VERSION bumped to 6.0.0.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Harden git clone attack surface for remote scans with defense-in-depth:
Layer 1 (all platforms): 8 git config flags disable hooks, symlinks,
filter/smudge drivers, fsmonitor, local file protocol. 4 env vars
isolate from system/user git config and block interactive prompts.
Layer 2 (OS sandbox): macOS sandbox-exec and Linux bubblewrap (bwrap)
restrict file writes to only the specific temp directory. bwrap
probe-tests availability before use. Graceful fallback on Windows
and Ubuntu 24.04+ (git config hardening only).
Additional: post-clone 100MB size check, UUID-unique evidence filenames,
evidence file cleanup, cleanup guarantee in scan/plugin-audit commands.
32 new tests (1147 total).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
