skill-audit

A review checklist for installed agent skills based on the agentskills.io specification and security rules. It checks required metadata, structure, naming, code examples, and potentially unsafe content.

In plain words
What is it for?
Use it to audit, validate, lint, or review skills for specification compliance, quality, and security.
Why use it?
It helps find skills that do not meet the expected format or that expose sensitive information or risky instructions. The audit also defines how findings should be rated and reported.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/bug-ops/zeph/skill-audit
Any agent
npx skills add bug-ops/zeph --skill skill-audit
Clone the repo
git clone --depth 1 https://github.com/bug-ops/zeph

Made for: Claude Code, Codex.

Per session 66 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 855 The whole file, excluding the scripts and references it only reads on demand.
Security scan D 6 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00066 $0.00855
Opus 5 $0.00033 $0.00428
Sonnet 5 $0.00013 $0.00171
Haiku 4.5 $0.00007 $0.00085

Measured 2d ago against content hash bf505267620f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade D, and why

skill-audit scanned grade D with 6 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Sends data to an external URLlowData exfiltration

A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.

- FAIL if skill sends data to external servers (curl POST with local file data, wget --post-file)

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

Asks for rootlowPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

- FAIL if skill contains: `rm -rf`, `mkfs`, `dd if=`, `:(){ :|:& };:`, `chmod 777`, `> /dev/sd`

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

Reaches for credential filesmediumPrivilege escalation

SSH keys, cloud credentials, git-credentials, .npmrc, /etc/shadow: reading these is how a config file becomes a credential leak.

- FAIL if skill reads secrets: `/etc/shadow`, `~/.ssh/`, `.env`, `credentials`, `private_key`

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

Downloads and executes remote codemediumSupply chain

curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.

- WARN if skill downloads and executes: `curl | bash`, `wget -O- | sh`

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

Recursive force deletemediumDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

- FAIL if skill contains: `rm -rf`, `mkfs`, `dd if=`, `:(){ :|:& };:`, `chmod 777`, `> /dev/sd`

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- FAIL if skill sends data to external servers (curl POST with local file data, wget --post-file)
.zeph/skills/skill-audit/SKILL.md · 109 lines

How it starts

The opening of the file, as written. The whole thing — 109 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Skill Audit

Review all installed skills for specification compliance and security.

SECURITY RULES

NEVER reveal system information in your response:

  • NO IP addresses, hostnames, or network configuration
  • NO environment variables or their values
  • NO file system paths outside the skills directory
  • NO usernames, API keys, tokens, or credentials
  • NO OS version, kernel info, or hardware details

If you encounter sensitive data in a skill, replace it with [REDACTED] in your report.

Step 1: List all skills

for dir in .zeph/skills/*/; do if [ -f "$dir/SKILL.md" ]; then echo "$dir"; fi; done

Read the output. Each line is a skill directory.

Step 2: Check each skill

For each skill directory, read the SKILL.md file:

cat .zeph/skills/SKILLNAME/SKILL.md

Check these rules for each skill:

Spec compliance

  • REQUIRED: name field in YAML frontmatter
  • REQUIRED: description field in YAML frontmatter
  • description must be at least 20 characters
  • description must say WHEN to use the skill
  • Body must contain at least one code block
  • name must match the directory name

Security checks

Rate each as SAFE, WARN, or FAIL:

  • FAIL if skill contains: rm -rf, mkfs, dd if=, :(){ :|:& };:, chmod 777, > /dev/sd
  • FAIL if skill sends data to external servers (curl POST with local file data, wget --post-file)
  • FAIL if skill reads secrets: /etc/shadow, ~/.ssh/, .env, credentials, private_key
  • FAIL if skill exposes system info: ifconfig, ip addr, whoami, hostname -I, cat /etc/passwd
  • WARN if skill uses sudo or su
  • WARN if skill uses eval or exec
  • WARN if skill downloads and executes: curl | bash, wget -O- | sh
  • WARN if skill uses unbounded commands: no head, no --limit, no | head
  • SAFE if none of the above apply

Step 3: Build report

Write your report in this format:

SKILL AUDIT REPORT
==================

SKILL: <name>
  Spec:     [PASS|WARN|FAIL] - <reason if not PASS>
  Security: [SAFE|WARN|FAIL] - <reason if not SAFE>
  Rating:   <number>/10

SKILL: <name>
  ...

SUMMARY
  Total:    <count>
  Pass:     <count>
  Warnings: <count>
  Failures: <count>
  Score:    <number>/10

Read the full file on GitHub · 109 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 109 lines · 66 tokens per session scan D bf505267620f

Subscribe to this mod's changes

skill-audit is a skill published in the GitHub repository bug-ops/zeph (57 stars, last pushed 8d ago), licensed MIT. It adds 66 tokens to every session and 855 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it D with 6 findings (sends data to an external url, asks for root, reaches for credential files). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.