agentleFS
mtarcuremtarcure/claude-vibe-squad/.agents/skills/dependency-health-triage/SKILL.md

dependency-health-triage

Skill220 starsChanged 20 days agoMITView on GitHub

Use when a lockfile or vulnerability-scanner dump contains many dependency alerts and you must decide what actually needs action: derive the resolved transitive graph, merge advisories, prove vulnerable-symbol reachability and preconditions, rank intrinsic impact, assess maintenance risk, and name fixes or accepted-risk revisit conditions. Not for one vendored codebase's patch-parity review.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run the scanners available on the host (osv-scanner, trivy, npm audit, pip-audit) and merge results by package and version, deduplicating …

    line 13
  2. Distinctive:

    Check exploitation preconditions the advisory assumes — a specific configuration, a parser mode, an exposed listener — against how this …

    line 15
  3. Distinctive:

    Every retained finding states its reachability and the calling path.

    line 23

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Dependency Health Triage

Turn a raw vulnerability-scanner dump into a short, ranked list of dependencies that actually need action.

Steps

  1. Produce the true dependency graph, direct and transitive, from the lockfile rather than the manifest; the manifest understates what ships.
  2. Run the scanners available on the host (osv-scanner, trivy, npm audit, pip-audit) and merge results by package and version, deduplicating advisories that describe the same defect.
  3. For each advisory, determine reachability: is the vulnerable symbol called, on which path, and with what input? An unreached CVE is inventory, not risk.
  4. Check exploitation preconditions the advisory assumes — a specific configuration, a parser mode, an exposed listener — against how this project actually uses the package.
  5. Rank surviving items with review-severity-ladder, using intrinsic impact, not CVSS alone; CVSS is context-free and systematically overstates.
  6. Assess maintenance health separately from vulnerabilities: release cadence, single-maintainer risk, unmaintained transitive pins, and packages whose upstream has been renamed or transferred.
  7. For each actionable item, name the fix — version bump, pin, patch, replace, or accept — and the compatibility risk of that fix.
  8. Record accepted risks explicitly with the reason and a revisit condition, so acceptance is a decision rather than a silence.

Acceptance

  • Findings come from the lockfile graph, with transitive dependencies included.
  • Every retained finding states its reachability and the calling path.
  • Unreachable and precondition-unmet advisories are listed as suppressed, with the reason.
  • Ranking uses intrinsic impact; raw CVSS is not the sole justification.
  • Each action names the fix and its compatibility risk; accepted risks state a revisit condition.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.