skill-audit: The First Tool to Manage Your Agent Skills Ecosystem

Created on July 20, 2026

2026   ·   ai-agents   open-source   cli   developer-tools   python   skills   agent-ecosystem   ·   ai   open-source   tools

The agent skills ecosystem is exploding. 77K stars on agent-skills alone (and gaining 1,000+ a day). Claude Code, Codex CLI, Cursor, Gemini CLI — every major agent platform now uses SKILL.md files to define reusable capabilities.

But there’s a problem: zero tooling to manage them.

You can’t audit them. You can’t validate them. You can’t see what’s installed, check for conflicts, or diff two sets of skills. When you have 50+ skills from different sources, it’s a mess waiting to happen.

Enter skill-audit

skill-audit is a Python CLI that does exactly what the name says — it audits your skills directory, tells you what’s valid, what’s broken, what conflicts, and what changed.

It’s not a platform. It’s not a framework. It’s a dead-simple CLI you run against any directory of SKILL.md files.

Three commands

skill-audit audit ./skills

Scans every SKILL.md file in a directory and validates:

  • YAML frontmatter — does it have the --- delimiters?
  • Required fields — name and description must be present
  • Semver versions — version should be valid semver if present
  • Broken references — related_skills and linked_files paths that don’t exist
  • Duplicate names — same skill name in two different files
  • Similar descriptions — two skills describing the same thing (edit distance ≤ 10)
$ skill-audit audit ~/.hermes/skills
Audit result for /root/.hermes/skills: PASS
Skills parsed: 82
Errors: 0
Warnings: 2

skill-audit inventory ./skills

Lists everything installed in a human-readable tree or machine-readable JSON:

$ skill-audit inventory ~/.hermes/skills
Skills
├── airtable | Airtable REST API via curl. Records CRUD, filters, upserts.
├── arxiv | Search arXiv papers by keyword, author, category, or ID.
├── ascii-art | ASCII art: pyfiglet, cowsay, boxes, image-to-ascii.
├── blogwatcher | Monitor blogs and RSS/Atom feeds via blogwatcher-cli tool.
├── chrome-cdp | Control Chrome via CDP — zero X server load.
...

skill-audit diff ./old-skills ./new-skills

Shows what changed between two skill directories — added, removed, or modified skills with version bumps.

Why I built this

I have 82 skills installed in my Hermes Agent setup. Eighty-two. And I’m not a power user — that’s just the default set plus what I’ve built for my daily workflow.

When I looked for a tool to tell me what was actually installed, whether they were valid, and whether anything conflicted — there was nothing. Not one tool. The entire ecosystem has 80K+ GitHub stars worth of skills content and zero management tooling.

That’s the gap.

The tech

Pure Python, zero heavy dependencies:

  • CLI — argparse (stdlib)
  • Frontmatter parsing — PyYAML
  • Output — ANSI-colored terminal or --json for piped workflows

The validator checks semantic versioning, detects broken cross-references, and even catches description similarity using edit distance. It’s designed to work with any agent ecosystem — Claude Code, Codex CLI, Cursor, Gemini CLI, Hermes Agent, or custom SKILL.md directories.

Try it

pip install skill-audit

# Validate your skills
skill-audit audit ./my-skills

# See what's installed
skill-audit inventory ./my-skills --json

# See what changed after an update
skill-audit diff ./old-skills ./new-skills

The repo is on GitHub: github.com/vikasudasi/skill-audit

What’s next

  • Remote inventory — audit skills from a GitHub repo URL, not just local paths
  • Skill linting — custom rules beyond the built-in checks (e.g. “no description longer than 200 chars”)
  • Format migration — convert between agent-specific skill formats
  • GitHub Action — auto-validate skills on every PR to a skills repo

This is the latest in my daily workflow: research a trend → identify a gap → build something → share it. Follow along on GitHub or LinkedIn.