Skip to content
Agentmart
Submit a tool

skill-audit:audit

Run a full skill activation audit experiment.

Skill by SeleznovIvan

IndexedMIT★ 1
Write a review

Chalk Labs

Need help deploying skill-audit:audit?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/seleznovivan-skill-audit-audit

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Run a full skill activation audit experiment. ALWAYS invoke this skill when the user says "skill audit", "audit skills", "test skill activation", or "run activation experiment". Do not attempt to test skill activation directly — use this skill first.

Use cases: testing, security

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/skill-audit-audit

What's inside

Skill name
skill-audit:audit
Description
Run a full skill activation audit experiment. ALWAYS invoke this skill when the user says "skill audit", "audit skills", "test skill activation", or "run activation experiment". Do not attempt to test skill activation directly — use this skill first.
License
MIT
Instructions
295 lines in SKILL.md · view on GitHub

Agent Skills spec issues

  • name: only lowercase letters, digits and single hyphens allowed
  • name: "skill-audit:audit" does not match folder "audit"

Files (1)

  • SKILL.md

SKILL.md (first lines)

# /skill-audit:audit — Skill Activation Experiment

Run a complete skill activation audit in 4 phases. Present results after each phase and ask the user before proceeding to the next.

## Allowed Tools

Read, Write, Edit, Glob, Grep, Bash, Task, AskUserQuestion,
mcp__cclogviewer__list_sessions, mcp__cclogviewer__list_projects,
mcp__cclogviewer__search_logs, mcp__cclogviewer__get_session_timeline,
mcp__cclogviewer__get_tool_usage_stats, mcp__cclogviewer__get_session_stats

---

## Phase 1: Skill Inventory & Directive Audit

### 1.1 Discover skills

Use Glob to find all `.claude/skills/*/SKILL.md` files in the current project.

### 1.2 Parse each skill

For each SKILL.md, read the file and extract:
- `name` from frontmatter
- `description` from frontmatter
- `keywords` from frontmatter
- Full markdown body

### 1.3 Score directive compliance (0–3)

Score each skill's `description` field:

| Points | Criterion | Pattern Examples |
|--------|-----------|-----------------|
| +1 | Activation directive | "ALWAYS invoke", "MUST use", "ALWAYS use this skill" |
| +1 | Negative constraint | "Do not ... directly", "Never attempt to ... without", "Do not run ... directly" |
| +1 | Keyword triggers present | `keywords:` field has 5+ entries |

### 1.4 Present results

Display a table:

```
Phase 1: Skill Inventory & Directive Compliance

| # | Skill | Description (first 60 chars) | Dir. Score | Issues |
|---|-------|------------------------------|------------|--------|
| 1 | dockerfile-generator | Docker expert. ALWAYS invoke... | 3/3 | — |
| 2 | git-workflow | Git version control expert... | 2/3 | Missing keyword triggers |
```

For any skill scoring < 3, show a specific recommendation:
- Missing activation directive: "Add 'ALWAYS invoke this skill when...' to description"
- Missing negative constraint: "Add 'Do not ... directly — use this skill first' to description"
- Missing keyword triggers: "Add a `keywords:` field with 5+ relevant terms"

### 1.5 Ask to continue

Ask the user which skills to test:
- **All** — test every discovered skill
- **Select specific** — let user pick from the list
- **Fix first** — pause to fix directive issues before testing

---

## Phase 2: Query Mining

### 2.1 Mine queries from session logs

Use the cclogviewer MCP tools to find real user queries that should have triggered each skill:

1. Call `mcp__cclogviewer__list_projects` to find the current project
2. Call `mcp__cclogviewer__list_sessions` to get recent sessions (limit 50)
3. For each skill, call `mcp__cclogviewer__search_logs` with each keyword from the skill's `keywords` field
4. Cross-reference with `mcp__cclogviewer__get_tool_usage_stats` to find sessions where the Skill tool was NOT invoked (missed activations)
5. Extract the user's original query text from matched log entries

### 2.2 Generate contextual queries (if needed)

If fewer than 5 real queries found per skill:

1. Use Glob to find relevant project files (e.g., for a dockerfile skill, look for `Dockerfile*`, `docker-compose*`, `.dockerignore`)
2. Generate queries that reference actual project context:
   - "Write a Dockerfile for the Express app in src/server/"
   - "Help me resolve the merge conflict in package.json"
3. Ensure queries are specific enough to avoid clarification responses
4. Tag generated queries as `generated`

### 2.3 Present query list

For each skill, show the queries:

```
Phase 2: Query Mining

dockerfile-generator (5 queries):
  [log] "help me write a dockerfile for this project"
  [log] "containerize the api server"
  [gen] "Write a Dockerfile for the Express app in src/server/"
  [gen] "Create a multi-stage Docker build for production"
  [gen] "Add a docker-compose.yml for local development"

git-workflow (4 queries):
  [log] "resolve the merge conflict in main"
  [log] "squash my last 3 commits"
  [gen] "Help me rebase feature-auth onto main"
  [gen] "Undo the last commit without losing changes"
```

### 2.4 Ask to continue

Ask the user:
- **Run inline** — execute trials now in this session
- **Generate script** — write a `run-trials.sh` script for later execution
- **Edit queries** — let user modify the query list first

---

## Phase 3: Activation Experiment

From SeleznovIvan/claude-skills-test, MIT.

Signals

Stars
1
Forks
0
Open issues
0
Last commit (this folder)
Mar 5, 2026 (8 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from SeleznovIvan

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get skill-audit:audit running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying skill-audit:audit?

CallGet help
Cite this page

skill-audit:audit. Agent Mart. https://agentmart.dev/skills/seleznovivan-skill-audit-audit (last updated 2026-10-24)