A minimal implementation of agent skills - a lightweight, open format for extending AI agents with specialized capabilities.
Overview
A minimal implementation of agent skills - a lightweight, open format for extending AI agents with specialized capabilities. Agent skills are markdown files (SKILL.md) that give AI agents specialized knowledge and workflows. Each skill contains: - : Name, description, metadata - : Detailed guidelines for the agent When a user's task matches a skill, the agent automatically activates it to provide expert assistance. - : Reviews code for bugs, security issues, and best practices - : Git workflows, commit messages, and troubleshooting - : Intelligent file organization strategies - : Makes actual HTTP requests to APIs and analyzes responses (GET, POST, PUT, PATCH, DELETE) 1. : Scans skills/ directory for SKILL.md files 2. : Parses YAML frontmatter (name, description) 3. : When user's task matches, loads full skill instructions 4. : Agent follows skill guidelines to complete task That's it! The skill will be automatically discovered.
README
Agent Skills
A minimal implementation of agent skills - a lightweight, open format for extending AI agents with specialized capabilities.
Written by Aman, Claude, and Cursor
What Are Agent Skills?
Agent skills are markdown files (SKILL.md) that give AI agents specialized knowledge and workflows. Each skill contains:
- YAML frontmatter: Name, description, metadata
- Markdown instructions: Detailed guidelines for the agent
When a user’s task matches a skill, the agent automatically activates it to provide expert assistance.
Quick Start
# Install
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
# Configure
cp .env.example .env
# Add your OPENAI_API_KEY to .env
# Run
python -m agent.chat
Or use the run script:
./run.sh
Included Skills
- code-review: Reviews code for bugs, security issues, and best practices
- git-helper: Git workflows, commit messages, and troubleshooting
- file-organizer: Intelligent file organization strategies
- api-tester: Makes actual HTTP requests to APIs and analyzes responses (GET, POST, PUT, PATCH, DELETE)
How It Works
- Discovery: Scans
skills/directory forSKILL.mdfiles - Loading: Parses YAML frontmatter (name, description)
- Activation: When user’s task matches, loads full skill instructions
- Execution: Agent follows skill guidelines to complete task
Creating Your Own Skill
mkdir -p skills/my-skill
Create skills/my-skill/SKILL.md:
---
name: my-skill
description: What this skill does
version: 1.0.0
---
# My Skill
You are an expert at [skill domain].
## Guidelines
1. Step one
2. Step two
3. Step three
## Examples
Show examples of using this skill effectively.
That’s it! The skill will be automatically discovered.
Project Structure
agent-skills/
├── agent/
│ ├── skills_manager.py # Skill discovery & loading
│ ├── llm_client.py # OpenAI client
│ ├── chat.py # Terminal interface
│ └── code_executor.py # Code execution for skills
├── skills/ # Skills directory
│ ├── code-review/
│ ├── git-helper/
│ ├── file-organizer/
│ └── api-tester/
├── requirements.txt
└── run.sh
Resources
- AgentSkills.io - Official standard and documentation
- Claude Skills - Anthropic’s skill examples
Learn More
Tutorial: Building Agent Skills from Scratch - Detailed guide on how this implementation works
Repository: github.com/onlyoneaman/agent-skills
License
MIT
Recommended Tools
Try a different keyword or remove a filter.
Install
npx skillfish add onlyoneaman/agent-skills