Agent Skills

A collection of reusable agent skills for Claude Code and other AI coding assistants. Each skill is a self-contained module with scripts, references, and configuration that extend an AI agent's capabilities.

Quick Start

Option 1: Copy individual skill

# Copy a skill into your project
cp -r skills/<skill-name> /path/to/your/project/.claude/skills/

# Configure environment
cd /path/to/your/project/.claude/skills/<skill-name>
cp .env.example .env
# Edit .env with your credentials

Option 2: Git submodule (all skills)

git submodule add https://github.com/user/agent-skills.git .claude/agent-skills

Then symlink or copy the skills you need into .claude/skills/.

Available Skills

Skill	Description	Tools
gitlab-self-hosted	GitLab Server MR workflow via REST API v4 - get/create MRs, assign reviewers, post reviews, analyze changes, staging merge, test case generation	curl, jq

Skill Structure

Each skill follows a consistent structure:

skills/<skill-name>/
├── SKILL.md          # Metadata (YAML frontmatter) + argument handling instructions
├── README.md         # Human-readable setup guide and usage docs
├── .env.example      # Environment variables template (copy to .env)
├── scripts/          # Executable bash/python scripts
└── references/       # Reference docs consumed by the AI agent at runtime

Key Files

• SKILL.md - The entry point for AI agents. Contains YAML frontmatter (name, description, version, license, allowed-tools) and argument routing table.
• README.md - Setup instructions, usage examples, and workflow documentation for humans.
• .env.example - Template for required environment variables. Never commit .env files.
• scripts/ - Self-contained scripts that handle API calls, data processing, etc.
• references/ - Markdown docs that the AI agent reads for context (API specs, workflows, best practices).

Contributing

1. Create a new directory under skills/ with a descriptive kebab-case name
2. Add SKILL.md with YAML frontmatter:

   ---
   name: my-skill
   description: What this skill does
   version: 1.0.0
   license: MIT
   allowed-tools:
     - Bash
   ---

3. Add README.md with setup and usage instructions
4. Add .env.example if the skill requires credentials or configuration
5. Add executable scripts in scripts/ and reference docs in references/
6. Ensure .env is in .gitignore (already configured at repo root)

Security

• Never commit .env files - they contain credentials and tokens
• The root .gitignore blocks all .env files by default
• Each skill provides .env.example as a safe template
• Use scoped tokens with minimal required permissions

License

MIT