Skill Audit — Claude Code skill for catalog checks
The Skill Audit Claude Code skill checks skill usage and whether descriptions match the prompts that trigger them. It also flags idle skills and recurring work with no matching skill.
MIT ★ 26 updated 2026-11-16
npx skills add wrg32786/aigent-os --skill skill-audit -a claude-codeWhat Skill Audit does
Skill Audit builds a report on a skill catalog. It sorts entries by their most recent invocation, counts calls over the last 90 days, and marks retirement candidates: skills unused for more than 60 days with fewer than three total calls.
It compares description triggers with prompts preceding each invocation. It also groups recent prompts that led to manual use of Read, Bash, or Grep instead of a skill; clusters with 3+ instances become possible capability gaps. The audit reads the skill index, skill folders, and session logs. Log locations vary by installation and may include vault/daily/, ~/.claude/projects/, or hook output logs.
The result is a Markdown report with four sections. An optional copy can go to vault/memory/SKILL_AUDIT_<YYYY-MM-DD>.md.
What it helps with
- find skills that have gone unused
- compare skill descriptions with prompts that trigger them
- spot recurring tasks handled with manual tool calls
- decide which skills to keep, archive, or delete
Example prompts for the agent
- Audit my skill catalog and produce a Markdown report.
- Find skills with little recent use and recommend whether to keep, archive, or delete them.
- Compare invocation prompts with skill descriptions and flag capability gaps.
How to install Skill Audit
1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill skill-audit -a claude-code2Manually: put the files in place
~/.claude/skills/skill-audit/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/skill-audit/
1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill skill-audit -a codex2Manually: put the files in place
.agents/skills/skill-audit/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill skill-audit -a cursor2Manually: put the files in place
.cursor/skills/skill-audit/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install wrg32786/aigent-os skills/skill-audit2Manually: put the files in place
.github/skills/skill-audit/SKILL.md1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill skill-audit -a gemini-cli2Platform CLI
gemini skills install https://github.com/wrg32786/aigent-os --path skills/skill-auditManually: put the files in place
.devin/skills/skill-audit/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill skill-audit -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about Skill Audit
How do I install Skill Audit in Claude Code?
Run npx skills add wrg32786/aigent-os --skill skill-audit -a claude-code. The skill file is at ~/.claude/skills/skill-audit/SKILL.md.
What does Skill Audit do?
It reports skill usage, prompt-to-description mismatches, and possible gaps where recurring work has no matching skill.
What data does Skill Audit inspect?
It reads the skill index and folders, invocation history, and session logs. Log locations depend on the installation and can include vault/daily/, ~/.claude/projects/, or hook output logs.
Source text
SKILL.md · MIT
--- name: Skill Audit description: "Audit skill usage and description-trigger fit. Use when finding unused skills, reviewing invocation history, or identifying missing capabilities." --- # Skill Audit Auditing the skill catalog itself. As the install grows from 5 skills to 50, the catalog goes from "I remember every skill" to "I have no idea what's in here." This skill makes the catalog legible to its owner. ## When to use - Once a quarter — hygiene pass - When the wrong skill is invoked repeatedly, or the right one never is (a description mismatch) - When a new feature feels like it should be a skill but you can't remember if you already built one - After importing a skill bundle from another principal (e.g., from a public aigent-OS release) - When considering retiring a project area and want to know which skills go with it ## What it produces A markdown report with four sections: ### Section 1 — All skills, sorted by last invocation ``` | Skill | Last used | Total invocations (90d) | Description | |---|---|---|---| | /open | today | 47 | Start session, load vault context | | /close | today | 47 | End session, commit memory | | /skill-recall | 5 days ago | 12 | Find the skill that fits a task | | /skill-name | 23 days ago | 3 | Example skill entry | | /old-experiment | 89 days ago | 1 | DEAD — flagged for retirement | ``` ### Section 2 — Dead skills (last used >60 days, <3 total invocations) These are candidates for retirement. List with: - Skill name + path - Last-used date - Total invocations - Recommendation: keep / archive / delete ### Section 3 — Description mismatch detection For each skill, compare the trigger words in its `description` (and `.claude/skill-index.json`, where kept) against the actual prompt patterns that preceded each invocation. Flag mismatches: ``` ⚠️ /channel-audit Triggers: "channel status", "YPP progress", "kill a channel" Actual prompts that led to invocation: "how are channels doing", "performance check", "review channels" Suggested words to ADD to the description: "performance check", "review channels", "how are channels doing" ``` ### Section 4 — Capability gaps Heuristic: scan recent prompts (last 30 days) where no skill was invoked and the principal manually invoked Read/Bash/Grep instead of a skill. Cluster by keywords. If a cluster has 3+ instances, flag as a capability gap. ``` 🆕 Possible missing skill: "weekly metrics rollup" Inferred from prompts: "weekly metrics", "this week's numbers", "rollup the week" Manual tools used: 4 Bash calls per occurrence Suggestion: codify as /weekly-rollup skill via /skills-builder ``` ## How to execute 1. Read the skill index and skill folder 2. Pull session logs / invocation history (location depends on install — check `vault/daily/`, `~/.claude/projects/`, hook output logs) 3. For each skill: count last 90 days of invocations, last invocation date 4. Build the four sections above 5. Output as markdown 6. Optionally: write report to `vault/memory/SKILL_AUDIT_<YYYY-MM-DD>.md` for trend tracking ## Why this skill exists Native skill matching only works if the catalog stays useful. Skills accumulate faster than they're retired. Trigger patterns drift from the principal's actual phrasing. Capability gaps go unnoticed for months. Skill audit is the corrective loop. ## Anti-patterns this prevents - Dead skills nobody has used in 6 months, spending the listing budget (noise) - Skills the principal needs going unmatched because their descriptions don't match how they actually phrase things - Re-building a skill that already exists (catalog isn't legible to the owner) - Frequently-manual workflows that should be skills but never get codified
© wrg32786 · MIT · wrg32786/aigent-os