Tell me about yours Skills

[japertechnology]

And how to add more

[github-actions]

Pi Skills

Skills are self-contained capability packages that pi loads on-demand. They provide specialized workflows, setup instructions, helper scripts, and reference documentation for specific tasks. Pi implements the Agent Skills standard.

How They Work

1. At startup, pi scans skill locations and extracts names + descriptions.
2. In the system prompt, available skills are listed so the agent knows what's available.
3. On demand, when a task matches a skill's description, the agent reads the full SKILL.md for detailed instructions.
4. Progressive disclosure — only short descriptions are always in context; full instructions load only when needed.

Skill Locations

Pi loads skills from these locations:

Location	Path
Global	~/.pi/agent/skills/
Project	.pi/skills/
Packages	skills/ dirs or pi.skills in package.json
Settings	skills array in settings.json
CLI	--skill <path> (repeatable)

Using a Skill

You can invoke skills with the /skill:name command:

/skill:brave-search              # Load and run
/skill:pdf-tools extract         # With arguments

---

How to Add a New Skill

1. Create a directory with your skill name (lowercase, hyphens allowed):

   my-skill/
   ├── SKILL.md          # Required
   ├── scripts/          # Optional helper scripts
   └── references/       # Optional docs
   

2. Write SKILL.md with required frontmatter:

   ---
   name: my-skill
   description: What this skill does and when to use it. Be specific!
   ---
   
   # My Skill
   
   ## Setup
   Run once: `npm install` or whatever is needed.
   
   ## Usage
   Instructions the agent follows when using this skill.

3. Place it in any skill location (e.g., ~/.pi/agent/skills/my-skill/ for global, or .pi/skills/my-skill/ for project-level).

4. That's it! Pi will discover it on next startup.

Key Rules for Skill Names

• 1–64 characters, lowercase a-z, 0-9, hyphens only
• No leading/trailing hyphens, no consecutive hyphens
• Directory name must match the name in frontmatter

Tips

• Description matters — it determines when the agent picks up the skill. Be specific (e.g., "Extracts text and tables from PDF files" not "Helps with PDFs").
• You can also use skills from Claude Code or Codex by pointing to their skill directories in settings.
• Check out existing skill repos for inspiration:
  • Anthropic Skills — document processing, web dev
  • Pi Skills — web search, browser automation, Google APIs

[japertechnology]

Write a great GitHub About text describing this repo

[github-actions]

Here's a great GitHub "About" description for this repo:

🦞 A drop-in AI agent that runs entirely through GitHub Issues and Actions. One folder, zero infrastructure — just open an issue and start chatting. Conversations are committed to git for persistent memory across sessions. Multi-provider LLM support (Anthropic, OpenAI, Gemini, xAI, and more).

Character count: ~290 (well within GitHub's 350-char limit for the About field)

If you want a shorter alternative:

🦞 Drop a single folder into any repo and get a fully functional AI agent — powered by GitHub Issues and Actions, with persistent memory via git. No servers, no infrastructure.

[japertechnology]

How big is .GITCLAW, just the code?

[github-actions]

Here's the breakdown of .GITCLAW (excluding node_modules, bun.lock, and the logo image):

Category	Size	Lines
TypeScript (actual code)	48 KB	667 lines
Markdown (docs, skills, prompts)	640 KB	13,137 lines
Config (JSON/YAML)	104 KB	—
Built JS	4 KB	—
Total	~2.4 MB	—

The core code is remarkably small — just 667 lines of TypeScript across 3 lifecycle files (GITCLAW-AGENT.ts, GITCLAW-INDICATOR.ts, GITCLAW-ENABLED.ts) plus the installer. The bulk of the folder is documentation and markdown-based configuration. The heavy lifting is delegated to the pi coding agent runtime pulled in via node_modules.