Publish Skills — Testing skill for Claude Code

publish-skills

Pre-publish checklist for [Agent Skills](https://agentskills.io) — the open standard for agent capabilities supported by [Claude Code](https://code.claude.com), [Cursor](https://cursor.com), [Gemini CLI](https://geminicli.com), and [many more](https://agentskills.io). Makes sure your skill is spec-compliant and GitHub-ready before you ship.

How skills work

Agent Skills are folders containing a `SKILL.md` file with YAML frontmatter and markdown instructions. When you install a skill into your agent's skills directory, the agent reads the frontmatter to decide when the skill is relevant, then loads the instructions on demand. No runtime, no build step — just structured markdown that agents know how to find and use.

What it does

Bridges the gap between "works locally" and "ready for open source." The Agent Skills spec only requires `name` + `description`, but most published skills ship with more. This checklist covers everything the spec doesn't tell you.

• Frontmatter audit — required, should-have, and optional fields
• Name validation — the exact rules (no consecutive hyphens, must match directory, etc.)
• Directory structure — what files to include and why
• README patterns — section-by-section guide based on well-received published skills
• Consistency review — 10-point checklist for catching pre-publish mistakes
• Common mistakes table — what goes wrong and how to detect it
• Sanitizer pipeline — for fleet skills with private context: strip internal paths, agent names, and war stories before publishing; defend with a deny-list scanner for credentials and residual leaks

Install

Claude Code

mkdir -p .claude/skills
cd .claude/skills
git clone https://github.com/jeremyknows/publish-skills.git

Cursor / Other agents

# Check your agent's docs for the skills directory location, then:
git clone https://github.com/jeremyknows/publish-skills.git

OpenClaw

cd ~/.openc