Skip to content
Back to skills

Dep Audit

ASecurity

Audit dependency security risks with an evidence-only verdict and an explicitly separate fix operation. Use for dependency vulnerability review. Not for: implicit upgrades, general code review, or treating a successful fix command as clearance. Output: audit findings, optional fix status, and a read-only security verdict.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 24, 2026
ai-agentssecurity

Works with

  • terminal

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned September 24, 2026

npx -y skills add hmj1026/dhpk --skill dep-audit --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Dep Audit?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Dep Audit
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/hmj1026-dep-audit/badge)](https://www.skillsdirectory.com/skills/hmj1026-dep-audit)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: dep-audit
description: 'Audit dependency security risks with an evidence-only verdict and an explicitly separate fix operation. Use for dependency vulnerability review. Not for: implicit upgrades, general code review, or treating a successful fix command as clearance. Output: audit findings, optional fix status, and a read-only security verdict.'
argument-hint: '[--level=<severity>] [--fix]'
metadata:
  dhpk-invocation-class: implicit-eligible
---

# Dependency Audit

## Workflow

Resolve `--level <severity>` to `low|moderate|high|critical`, defaulting to
`moderate`, and accept `--fix` only when explicitly supplied. Prefer the
consumer project's `$PROJECT_DIR/.claude/scripts/dep-audit.sh` (`$PROJECT_DIR`
means the explicitly selected consumer project root); if it exists, a failed run
is a real terminal audit failure and does not trigger fallback. Otherwise use
the ecosystem commands in [`references/workflow.md`](references/workflow.md).

Run the audit first. Only after a successful audit may explicit `--fix` run
the matching fix command; no fix is implicit. Finish with the separate,
read-only independent security evidence review defined in the local workflow.
An available `$change-verdict --mode security` is an optional way to obtain
that review; its installation is not required. The verdict may
inspect the audit output, but it never runs a fixer and a successful `--fix`
does not clear the original audit or prove a secure result.

## When NOT to Use

- Dependency changes are desired without an explicit `--fix`.
- Code, document, test, or merge review is the target.
- A security verdict is being inferred from an unavailable or failed audit.

## Output

Report ecosystem, level, command, exit status, severity counts, vulnerability
details, and whether an explicit fix was attempted. Keep the audit result and
the independent security result separate. End with `PASS`, `FAIL`,
`BLOCKED`, or `UNAVAILABLE`; unknown or missing evidence is not a pass.

## Verification

- [ ] The level and fix flag were preserved.
- [ ] Project script precedence and failure behavior were preserved.
- [ ] Audit completed before any explicit fix; no implicit mutation occurred.
- [ ] The final security verdict was read-only and did not treat fix success as
      clearance.

## References

- [`references/workflow.md`](references/workflow.md) — project-script
  precedence, ecosystem commands, fix boundary, and verdict output.

Files in this skill

  • SKILL.md2.4 KB
  • agents/openai.yaml195 B
  • references/workflow.md2.6 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.