Hex Dependency Audit

Non-mutating supply-chain audit for Hex packages. Runs an 8-rule MVP catalogue against changed packages, enriches with Hex API metadata, wraps existing tools (mix hex.audit, mix_audit, OSV-Scanner), and emits a triage table.

When to Use

• After mix deps.update or mix deps.get brought in new versions
• On PRs that touch mix.lock (pre-merge gate)
• Before manually updating a single package (--preview <pkg>)
• When investigating a dependency you don't recognize

Iron Laws

1. NEVER claim a diff is clean without inspecting it. Run all 8 rules on the unpacked NEW tarball. "Looks fine" without a tool run is a false pass. Always write .claude/deps-audit/last-run.json — its absence is evidence the audit didn't actually run.
2. NEVER install mix_audit / osv-scanner — even if asked. Detect, warn with install instructions, skip cleanly if missing. If the user says "install it," respond with the install command (e.g., mix deps.add mix_audit --only dev) and do not execute it. The audit skill is non-mutating; mix.exs / mix.lock are off-limits regardless of consent.
3. NEVER promote a finding to BLOCK without rule citation. Every finding shows rule_id, severity, file:line, snippet, message. No handwaving.
4. NEVER fetch from Hex API without rate-limiting. Cap at 5 req/sec. Cache metadata 7 days, top-500 list 1 day.
5. NEVER run the audit on already-committed lock changes silently — tell the user which mode (A/B/C) is active and which (old, new) pairs resolved.
6. LLM triage only above threshold. Native rules + Semgrep + YARA are deterministic. The hex-deps-triager agent runs only when score

   10 (1 BLOCK or 3+ WARNs), and its verdicts are advisory — never auto-suppress a finding without human review.

Operating Modes

See ${CLAUDE_SKILL_DIR}/references/operating-modes.md for full resolver logic.

Execution Flow

Default = full 8-rule scan with streaming progress. --quick opts out to CVE + retirement only. See ${CLAUDE_SKILL_DIR}/references/execution-flow.md.

Step 1: Resolve the diff

Parse the mix.lock Erlang term format for both old and new sources. Emit a list of {pkg, old_version, new_version} tuples. Surface new-only and removed-only packages separately (a removed package is not audited; a brand-new package gets old_version = nil and skips diff-only rules).

See ${CLAUDE_SKILL_DIR}/references/diff-resolver.md for shell + mix run -e snippets per mode and the JSON output contract.

Step 2: Fetch tarballs (per-run tmpdir)

For each (pkg, old, new):

mix hex.package fetch <pkg> <old> --unpack -o ${AUDIT_TMPDIR}/tarballs/<pkg>/<old>/
mix hex.package fetch <pkg> <new> --unpack -o ${AUDIT_TMPDIR}/tarballs/<pkg>/<new>/

All ephemeral artifacts live under ${AUDIT_TMPDIR} (driver-owned, removed on exit). See ${CLAUDE_SKILL_DIR}/references/audit-tmpdir.md and ${CLAUDE_SKILL_DIR}/references/tarball-fetcher.md.

Step 3: Run the 8 MVP rules on each NEW tarball

Full catalogue (35 rules, MVP marked) in ${CLAUDE_SKILL_DIR}/references/heuristics.md. Bash + mix run -e implementations for all 8 MVP rules in ${CLAUDE_SKILL_DIR}/references/rules-impl.md (single-pass NEW + diff rules + Hex API rules, with run_all_rules master loop).

Step 4: External tool wrappers (parallel)

• mix hex.audit — retired-package check, always available
• mix_audit — CVE check via GHSA, if installed (else warn + skip; do NOT install)
• osv-scanner — CVE check via OSV.dev, if installed (else warn + skip; do NOT install)

See ${CLAUDE_SKILL_DIR}/references/external-tools.md for detection, output parsing, and severity mapping per tool.

Step 5: Hex API enrichment (per package)

• GET /api/packages/:name — owners, downloads, inserted_at
• GET /api/packages/:name/releases/:version — per-release publisher
• Compute: days_since_publish, owner_age_days, download_velocity

Cap at 5 req/sec. Per-run cache under ${AUDIT_TMPDIR}/hex-api/. See ${CLAUDE_SKILL_DIR}/references/hex-api.md for endpoint contracts, caching strategy, Rule 6/8 detection, and Levenshtein implementation.

Step 5.5: Apply hex_vet.exs ledger (if present)

If hex_vet.exs exists at project root, vetted-version findings are downgraded to INFO. Unvetted versions retain their severity. Lock-vs-ledger disagreement: lock wins. See the deps-vet skill's hex-vet schema doc for the "Lock-vs-ledger disagreement" section.

Use /phx:deps-vet <pkg> <version> (separate skill) to add entries.

Step 5.7: Differential subtract

When run with DIFFERENTIAL=1 (default), findings that existed in the OLD tarball are downgraded to INFO. Net-new signals reach the renderer at full severity. See ${CLAUDE_SKILL_DIR}/references/differential.md.

Step 5.8: LLM triage (when score > threshold)

For packages where the aggregate score exceeds 10, the hex-deps-triager sonnet agent reads finding + diff windows and produces structured verdicts (confidence, verdict, rationale, fp_reasons[]). A context-supervisor (haiku) consolidates verdicts across packages into triage/consolidated.md. Main skill reads only the consolidated file.

See ${CLAUDE_SKILL_DIR}/references/llm-triage.md.

Step 6: Score & render

Per-package weighted sum: BLOCK = 10, WARN = 3, INFO = 1. Risk band: 0 clean · 1–5 low · 6–15 medium · 16+ high.

Output:

1. Stdout: markdown table — pkg | old → new | risk | findings | diff.hex.pm | maintainer-change plus a per-package detail section for any non-clean row.
2. Sidecar (MANDATORY): Write .claude/deps-audit/last-run.json. The Phase 3 gate reads this; an audit that doesn't write it is a no-op for the gate. Always emit, even on clean runs.

--json flag emits JSON to stdout instead of markdown. See ${CLAUDE_SKILL_DIR}/references/output-renderer.md for table format, sidecar schema, exit-code rubric, and --quiet mode.

Out of scope / Phase 3 surface

• NEVER modify mix.lock, mix.exs, or any project file (non-mutating)
• NEVER auto-install missing tools (warn + skip)
• Gate mix deps.{get,update,compile} via deps-audit-gate.sh. See references/hook.md.
• Prompt for /phx:compound after BLOCK findings — corpus self-feeds.
• Emit SARIF 2.1.0 via --sarif <path> and gate CI via --ci.

References

• ${CLAUDE_SKILL_DIR}/references/heuristics.md — full 35-rule catalogue
• ${CLAUDE_SKILL_DIR}/references/rules-impl.md — bash + mix run -e for the 8 MVP rules
• ${CLAUDE_SKILL_DIR}/references/operating-modes.md — Mode A/B/C resolver
• ${CLAUDE_SKILL_DIR}/references/diff-resolver.md — shell snippets, lock parser
• ${CLAUDE_SKILL_DIR}/references/tarball-fetcher.md — fetch wrapper, parallel cap, cache prune
• ${CLAUDE_SKILL_DIR}/references/external-tools.md — mix_audit,