Awesome Agent Skills

A curated collection of high-quality, production-ready skills designed to supercharge AI coding agents — particularly Qwen Code, Claude Code, and other agent-based development workflows.

Each skill is a self-contained, modular package that extends your AI agent's capabilities with structured workflows, best practices, and domain-specific expertise.

---

📦 Available Skills

🧠 Brainstorming

Turn raw ideas into validated designs before writing a single line of code.

This skill transforms vague concepts into clear, actionable specifications through disciplined reasoning and collaborative dialogue. It prevents premature implementation, hidden assumptions, and misaligned solutions by enforcing a structured design process.

Key features:

• One-question-at-a-time discovery process
• Mandatory non-functional requirements analysis
• Understanding Lock — a hard gate before any design proposal
• 2–3 approach exploration with explicit trade-off analysis
• Incremental design presentation (200–300 words per section)
• Decision Log maintenance for documentation
• Hard exit criteria before implementation handoff

Best for: Architecture planning, feature design, system behavior specification, and any high-impact decision-making.

---

✍️ Prompt Engineering Patterns

Master advanced prompt engineering techniques to maximize LLM performance, reliability, and controllability.

A comprehensive guide to building production-grade prompts with proven patterns, from few-shot learning to chain-of-thought reasoning and template systems.

Key features:

• Few-shot learning with semantic similarity selection
• Chain-of-thought prompting (zero-shot, few-shot, self-consistency)
• Iterative prompt optimization with A/B testing
• Modular template systems with variable interpolation
• System prompt design for specialized assistants
• RAG system integration patterns
• Token efficiency and latency reduction strategies
• Performance metrics tracking (accuracy, consistency, latency)

Best for: Building complex LLM applications, optimizing prompt reliability, designing AI assistant behaviors, and creating reusable prompt templates.

---

🐍 Python Pro

Master Python 3.12+ with modern features, async patterns, performance optimization, and production-ready practices.

Expert-level guidance on the latest Python ecosystem, including uv, ruff, pydantic, FastAPI, and the full spectrum of modern Python development.

Key capabilities:

• Modern Python: 3.12+ features, advanced async/await, pattern matching, type hints & generics
• Tooling: uv package management, ruff linting/formatting, mypy/pyright type checking
• Testing: pytest, property-based testing with Hypothesis, coverage analysis, benchmarking
• Performance: cProfile, py-spy, async optimization, multiprocessing, caching strategies
• Web Development: FastAPI, Django, Flask, Pydantic, SQLAlchemy 2.0+ with async
• Data & ML: NumPy, Pandas, Scikit-learn, visualization, ETL pipelines
• DevOps: Docker, Kubernetes, cloud deployment, CI/CD, monitoring

Best for: Writing production-grade Python, migrating to modern tooling, optimizing async/performance, designing APIs, and building data pipelines.

---

🛠️ Skill Creator

Create, test, and iteratively improve agent skills with a rigorous evaluation framework.

A meta-skill for building and refining skills themselves. Provides a complete co-development loop: draft → test → evaluate → improve, with quantitative benchmarking and a visual review interface.

Key features:

• Structured skill anatomy with progressive disclosure
• Test case generation with should/should-not trigger evals
• Parallel with-skill vs. baseline comparison runs
• Quantitative assertion drafting and automated grading
• Interactive eval viewer with qualitative feedback collection
• Description optimization loop for accurate skill triggering
• Blind comparison mode for rigorous A/B testing
• Benchmark aggregation with timing and token usage metrics

Best for: Creating new skills from scratch, improving existing skills, measuring skill performance, and optimizing skill descriptions for better triggering accuracy.

---

🎨 Tailwind Patterns

Tailwind CSS v4 principles with CSS-first configuration, container queries, and modern design token architecture.

Up-to-date guidance for the latest Tailwind CSS, embracing the Oxide engine, CSS-native configuration, and modern responsive design patterns.

Key features:

• CSS-first @theme configuration (no more tailwind.config.js)
• Container queries for component-level responsiveness
• OKLCH color system with semantic token architecture
• Mobile-first responsive design with modern breakpoint patterns
• Dark mode strategies (class, media, selector)
• Modern layout patterns: Flexbox, Grid, asymmetric/Bento layouts
• Component extraction signals and anti-patterns
• Performance optimization with the Oxide engine

Best for: Migrating to Tailwind v4, building responsive components, implementing design systems, and mastering modern CSS utility patterns.

---

🐚 Unix Philosophy

Apply UNIX philosophy principles to any coding or system design conversation.

A design advisor that guides you toward software that does one thing well, composes naturally with other programs, and stays simple enough to understand and change. Based on Eric Raymond's The Art of Unix Programming.

Original skill by Khaled Mahfouz with aid from Dr. Nibras Abu Alzahab.

Key features:

• All 17 UNIX philosophy principles with practical interpretations
• Context-aware advising (code review, new design, general questions, proactive flagging)
• Signal-based violation detection (god functions, mixed concerns, implicit state)
• Before/after code examples in multiple languages (Python, Go, Bash)
• Separation of policy and mechanism patterns
• Composition via standard streams guidance
• Design conversation facilitation framework
• Language-agnostic — applies to Python, Go, Bash, JS, Rust, etc.

Best for: Code reviews, architecture design, refactoring discussions, simplifying complex systems, and building composable, maintainable software.

---

🚀 How to Use These Skills

Skills can be installed globally (available in every project) or locally (scoped to a single project). Both Qwen Code and GitHub Copilot support custom skills with auto-discovery.

---

Qwen Code

Qwen Code discovers skills from two locations automatically. You can install any or all skills from this collection.

Global Installation (All Projects)

Skills placed in ~/.qwen/skills/ are available across every Qwen Code session, regardless of the project you're working on.

# Clone the repository
git clone https://github.com/fonte-aperta/awesome-agent-skills.git

# Copy all skills to the global directory
cp -r awesome-agent-skills/* ~/.qwen/skills/

# Or install individual skills
cp -r awesome-agent-skills/unix-philosophy ~/.qwen/skills/
cp -r awesome-agent-skills/python-pro ~/.qwen/skills/

Note: Launch Qwen Code with the --experimental-skills flag to enable skill discovery and triggering.

Local / Project Installation (Single Project)

Skills placed in .qwen/skills/ at your project root are only available within that specific project. This is ideal for team-shared skills or project-specific workflows.

# Inside your project directory
mkdir -p .qwen/skills

# Copy all skills
cp -r /path/to/awesome-agent-skills/* .qwen/skills/

# Or install individual skills
cp -r /path/to/awesome-agent-skills/tailwind-patterns .qwen/skills/

Scope	Path	Use When
Global	~/.qwen/skills/<skill-name>/SKILL.md	You want the skill everywhere
Local	.qwen/skills/<skill-name>/SKILL.md	The skill is project-specific

---

GitHub Copilot

GitHub Copilot supports the open Agent Skills standard. Skills are auto-discovered when placed in the correct directory and triggered based on their description field.

Global Installation (All Repos)

# Clone the repository
git clone https://github.com/fonte-aperta/awesome-agent-skills.git

# Copy all skills to the global Copilot skills directory
cp -r awesome-agent-skills/* ~/.copilot/skills/

# Or install individual skills
cp -r awesome-agent-skills/brainstorming ~/.copilot/skills/

Local / Project Installation (Single Repo)

Place skills inside .github/skills/ in your repository. These skills are committed to version control and shared with your team.

# Inside your repository
mkdir -p .github/skills

# Copy all skills
cp -r /path/to/awesome-agent-skills/* .github/skills/

# Or install individual skills
cp -r /path/to/awesome-agent-skills/python-pro .github/skills/

Commit .github/skills/ to share with your team — every collaborator gets the same agent capabilities.

Scope	Path	Use When
Global	~/.copilot/skills/<skill-name>/SKILL.md	Personal use across all repos
Local	.github/skills/<skill-name>/SKILL.md	Team-shared, repo-specific skills

---

Skill Structure

Each skill follows a standard layout:

skill-name/
├── SKILL.md              # Required — skill definition with frontmatter
├── references/           # Optional — detailed documentation loaded on demand
├── scripts/              # Optional — executable utilities for deterministic tasks
└── assets/               # Optional — templates, examples, static resources

Triggering

Skills are triggered automatically based on their description field in the YAML frontmatter. The description defines when the skill activates based on user input patterns, keywords, and context.

To use a skill explicitly, mention its purpose in your prompt:

• "Let's brainstorm the architecture for..." → Brainstorming
• "Help me optimize this prompt..." → Prompt Engineering Patterns
• "Write a production-ready Python service..." → Python Pro
• "Create a skill for..." → Skill Creator
• "Style this component with Tailwind..." → Tailwind Patterns
• "Review this code through a UNIX philosophy lens..." → Unix Philosophy

---

Quick Install All

Want everything at once?

# Qwen Code — Global
git clone https://github.com/fonte-aperta/awesome-agent-skills.git
cp -r awesome-agent-skills/* ~/.qwen/skills/

# GitHub Copilot — Global
cp -r awesome-agent-skills/* ~/.copilot/skills/

# GitHub Copilot — Local (in your repo)
cp -r awesome-agent-skills/* .github/skills/

---

📋 Skill Comparison

Skill	Domain	Complexity	Best For
Brainstorming	Architecture & Design	Medium	Planning before coding
Prompt Engineering	LLM Optimization	Medium	Building better prompts
Python Pro	Python Development	High	Production Python
Skill Creator	Meta (Skill Building)	High	Creating & testing skills
Tailwind Patterns	CSS / Frontend	Medium	Modern Tailwind v4
Unix Philosophy	Design Principles	Low–Medium	Code quality & architecture

---

🤝 Contributing

We welcome contributions! To add a new skill:

1. Use the Skill Creator skill to guide you through the process
2. Follow the standard skill structure
3. Include a clear description for triggering
4. Add tests/evals where applicable
5. Submit a pull request

See skill-creator/SKILL.md for detailed instructions on creating and evaluating skills.

---

📄 License

This project is licensed under