dmonteroh

deps-audit

"Produces a local, best-effort dependency audit summary and remediation plan for repos with dependency manifests."

dmonteroh 1 Updated 3mo ago

Resources

2
GitHub

Install

npx skillscat add dmonteroh/curated-agent-skills/deps-audit

Install via the SkillsCat registry.

SKILL.md

deps-audit

Provides a fast, low-noise dependency audit that relies on locally available tooling and produces an actionable remediation plan.

Use this skill when

  • A repo includes dependency manifests or lockfiles that need a security or license review.
  • A release needs a quick dependency risk assessment with actionable next steps.
  • A supply-chain or SBOM check is required using local tooling.

Do not use this skill when

  • No dependency manifests or lockfiles are available.
  • Only high-level strategy discussion is needed without producing an audit report.
  • The environment cannot write output files under the repo.

Required inputs

  • Target repo path (default: current working directory).
  • Any policy constraints (severity threshold, license allow/deny list).
  • Output location preference (default: docs/_docgen/deps-audit/).

Constraints

  • Uses local tooling only; no network access assumptions.
  • Does not install or modify dependencies.
  • Writes outputs only under the specified repo path.

Workflow

  1. Scope and detect manifests
  • Identify dependency manifests and lockfiles by ecosystem.
  • Output: list of detected ecosystems and evidence paths.
  1. Collect local scan evidence
  • Run best-effort local tools per ecosystem; do not install tools.
  • If a tool is missing, record the recommended command and continue.
  • Output: raw tool outputs under docs/_docgen/deps-audit/raw/.
  1. Analyze findings
  • Summarize vulnerabilities by severity, reachability, and exposure.
  • Summarize license and SBOM signals when available.
  • Output: prioritized findings table with affected packages and suggested fixes.
  1. Produce remediation plan
  • Propose minimal, compatible upgrades or mitigations.
  • Flag items needing architectural decisions.
  • Output: ordered remediation actions with effort notes.

Decision points

  • If no manifests are detected, stop and report "no dependency evidence found".
  • If scan tooling is missing, document the missing tool and skip that scan.
  • If license policy is provided, classify conflicts; otherwise list notable licenses only.
  • If only transitive vulnerabilities exist with no direct fix, recommend pinning or replacement options.

Common pitfalls

  • Ignoring lockfiles and using manifests only (reduces accuracy).
  • Treating dev-only dependencies as production impact without justification.
  • Performing broad major upgrades without assessing compatibility.
  • Assuming network access or installing tools without approval.

Output contract

Report a concise summary with the following sections:

  • Format: markdown with headings and bullet lists.
  • Detected ecosystems + evidence files.
  • Tools executed and missing tools (with suggested commands).
  • Top vulnerabilities (severity, package, version, reachability, fix).
  • License or SBOM findings and conflicts (if tooling available).
  • Remediation plan (ordered actions + notes).
  • Limitations or gaps (missing tools, missing manifests).

Scripts

scripts/deps.sh provides scan and report commands.

  • Usage:
    • ./skills/deps-audit/scripts/deps.sh scan
    • ./skills/deps-audit/scripts/deps.sh report
    • Optional: set DEPS_REPO_ROOT=/path/to/repo to target another repo.
  • Required tools (best-effort only): npm, pnpm, yarn, pip-audit, govulncheck, cargo-audit.
  • Verification:
    • Confirm docs/_docgen/deps-audit/REPORT.md exists.
    • Confirm raw outputs exist under docs/_docgen/deps-audit/raw/.

Examples

Example output snippet:

  • Detected ecosystems: node, python
  • Missing tools: pip-audit (suggested: pip-audit -f json)
  • Top vulnerabilities: package-x@1.2.3 (high, reachable)
  • Remediation: upgrade package-x to 1.2.9

References

  • references/README.md

Categories