Audits a Claude Code skill for security risks in three modes: before download (from a URL or install command), after download but before install (from a .skill file), or after install (from a local skills directory). Use this skill whenever a user is about to install a skill from any source — including GitHub URLs, git clone commands, npx/npm commands, curl/wget downloads, pip installs, marketplace links, or raw SKILL.md URLs. Also trigger when a user asks "is this skill safe?", "should I trust this skill?", "can you check this before I install it?", "audit this skill", or pastes any link to a skill repository or .skill file. If a user mentions installing ANY skill, proactively offer to audit it first — do not wait for them to ask.
97
97%
Does it follow best practices?
Impact
99%
1.28xAverage score across 5 eval scenarios
Advisory
Suggest reviewing before use
Use this template for every audit output. Fill in all sections. Omit sections that have no findings (e.g. if there are no CRITICAL issues, omit that block).
═══════════════════════════════════════════════
SKILL SAFETY AUDIT REPORT
═══════════════════════════════════════════════
Skill: [name from frontmatter, or URL if unnamed]
Source: [URL audited]
Audited on: [date]
Scripts found: [count] ([list filenames or "none"])
───────────────────────────────────────────────
OVERALL VERDICT
───────────────────────────────────────────────
[Choose one and remove the others]
🔴 DO NOT INSTALL — Critical issues found.
🟡 PROCEED WITH CAUTION — Warnings found. Review remedies below.
🟢 APPEARS SAFE — No significant issues detected. (See notes if any.)
───────────────────────────────────────────────
🔴 CRITICAL ISSUES ([count])
───────────────────────────────────────────────
[For each critical issue:]
[Check ID] — [Check Name]
Found in: [SKILL.md / script filename]
Detail: [Exact excerpt or pattern that triggered this check, quoted]
Why this matters: [1–2 sentences in plain language]
Action: Do not install this skill. [Any additional guidance e.g. report link]
───────────────────────────────────────────────
🟡 WARNINGS ([count])
───────────────────────────────────────────────
[For each warning:]
[Check ID] — [Check Name]
Found in: [SKILL.md / script filename]
Detail: [Exact excerpt or pattern that triggered this check, quoted]
Why this matters: [1–2 sentences in plain language]
Remedy: [Step-by-step what to check or do before proceeding]
───────────────────────────────────────────────
🟢 INFO / NOTES ([count])
───────────────────────────────────────────────
[For each info item:]
[Check ID] — [Check Name]
Note: [Brief plain-language note]
───────────────────────────────────────────────
WHAT WAS REVIEWED
───────────────────────────────────────────────
✅ SKILL.md frontmatter (allowed-tools, name, description)
✅ SKILL.md body (instructions, prompt injection patterns)
[✅ / ⚠️ not fetched] scripts/[filename] — [brief note if not fetched]
[repeat for each script]
───────────────────────────────────────────────
WHAT WAS NOT REVIEWED
───────────────────────────────────────────────
[List anything that could not be audited, e.g.:]
- Referenced scripts that returned 404
- Assets/binary files (not auditable via text review)
- Runtime behaviour (this audit is static analysis only)
───────────────────────────────────────────────
REMINDER
───────────────────────────────────────────────
This is a static pre-install review, not a guarantee of safety.
Even a clean audit does not protect against:
- Supply chain attacks (repo contents changed after audit)
- Runtime behaviour not visible in source
- Skills updated after you install them
When in doubt, don't install.
═══════════════════════════════════════════════| Condition | Verdict |
|---|---|
| Any 🔴 CRITICAL finding | DO NOT INSTALL |
| One or more 🟡 WARNINGs, no CRITICALs | PROCEED WITH CAUTION |
| Cannot fetch SKILL.md at all | DO NOT INSTALL (unverifiable) |
| No findings at any severity | APPEARS SAFE |
Always remind the user that a clean audit is not a guarantee.
audit-sample
evals
scenario-1
scenario-2
scenario-3
scenario-4
scenario-5