diff --git a/LICENSE b/LICENSE new file mode 100644 index 0000000..b3b379e --- /dev/null +++ b/LICENSE @@ -0,0 +1,120 @@ +Creative Commons Zero v1.0 Universal + +CREATIVE COMMONS CORPORATION IS NOT A LAW FIRM AND DOES NOT PROVIDE LEGAL SERVICES. +DISTRIBUTION OF THIS DOCUMENT DOES NOT CREATE AN ATTORNEY-CLIENT RELATIONSHIP. +CREATIVE COMMONS PROVIDES THIS INFORMATION ON AN "AS-IS" BASIS. CREATIVE COMMONS +MAKES NO WARRANTIES REGARDING THE USE OF THIS DOCUMENT OR THE INFORMATION OR +WORKS PROVIDED HEREUNDER, AND DISCLAIMS LIABILITY FOR DAMAGES RESULTING FROM +THE USE OF THIS DOCUMENT OR THE INFORMATION OR WORKS PROVIDED HEREUNDER. + +Statement of Purpose + +The laws of most jurisdictions throughout the world automatically confer +exclusive Copyright and Related Rights (defined below) upon the creator and +subsequent owner(s) (each and all, an "owner") of an original work of +authorship and/or a database (each, a "Work"). + +Certain owners wish to permanently relinquish those rights to a Work for the +purpose of contributing to a commons of creative, cultural and scientific +works ("Commons") that the public can reliably and without fear of later +claims of infringement build upon, modify, incorporate in other works, reuse +and redistribute as freely as possible in any form whatsoever and for any +purposes, including without limitation commercial purposes. These owners may +contribute to the Commons to promote the ideal of a free culture and the +further production of creative, cultural and scientific works, or to gain +reputation or greater distribution for their Work in part through the use and +efforts of others. + +For these and/or other purposes and motivations, and without any expectation +of additional consideration or compensation, the person associating CC0 with a +Work (the "Affirmer"), to the extent that he or she is an owner of Copyright +and Related Rights in the Work, voluntarily elects to apply CC0 to the Work +and publicly distribute the Work under its terms, with knowledge of his or her +Copyright and Related Rights in the Work and the meaning and intended legal +effect of CC0 on those rights. + +1. Copyright and Related Rights. + +A Work made available under CC0 may be protected by copyright and related or +neighboring rights ("Copyright and Related Rights"). Copyright and Related +Rights include, but are not limited to, the following: + + i. the right to reproduce, adapt, distribute, perform, display, + communicate, and translate a Work; + ii. moral rights retained by the original author(s) and/or performer(s); +iii. publicity and privacy rights pertaining to a person's image or likeness + depicted in a Work; + iv. rights protecting against unfair competition in regards to a Work, + subject to the limitations in paragraph 4(a), below; + v. rights protecting the extraction, dissemination, use and reuse of data + in a Work; + vi. database rights (such as those arising under Directive 96/9/EC of the + European Parliament and of the Council of 11 March 1996 on the legal + protection of databases, and under any national implementation thereof, + including any amended or successor version of such directive); and +vii. other similar, equivalent or corresponding rights throughout the world + based on applicable law or treaty, and any national implementations + thereof. + +2. Waiver. + +To the greatest extent permitted by, but not in contravention of, applicable +law, Affirmer hereby overtly, fully, permanently, irrevocably and +unconditionally waives, abandons, and surrenders all of Affirmer's Copyright +and Related Rights and associated claims and causes of action, whether now +known or unknown (including existing as well as future claims and causes of +action), in the Work (i) in all territories worldwide, (ii) for the maximum +duration provided by applicable law or treaty (including future time +extensions), (iii) in any current or future medium and for any number of +copies, and (iv) for any purpose whatsoever, including without limitation +commercial, advertising or promotional purposes (the "Waiver"). Affirmer makes +the Waiver for the benefit of each member of the public at large and to the +detriment of Affirmer's heirs and successors, fully intending that such Waiver +shall not be subject to revocation, rescission, cancellation, termination, or +any other legal or equitable action to disrupt the quiet enjoyment of the Work +by the public as contemplated by Affirmer's express Statement of Purpose. + +3. Public License Fallback. + +Should any part of the Waiver for any reason be judged legally invalid or +ineffective under applicable law, then the Waiver shall be preserved to the +maximum extent permitted taking into account Affirmer's express Statement of +Purpose. In addition, to the extent the Waiver is so judged Affirmer hereby +grants to each affected person a royalty-free, non transferable, non +sublicensable, non exclusive, irrevocable and unconditional license to +exercise Affirmer's Copyright and Related Rights in the Work (i) in all +territories worldwide, (ii) for the maximum duration provided by applicable +law or treaty (including future time extensions), (iii) in any current or +future medium and for any number of copies, and (iv) for any purpose +whatsoever, including without limitation commercial, advertising or +promotional purposes (the "License"). The License shall be deemed effective +as of the date CC0 was applied by Affirmer to the Work. Should any part of the +License for any reason be judged legally invalid or ineffective under +applicable law, such partial invalidity or ineffectiveness shall not +invalidate the remainder of the License, and in such case Affirmer hereby +affirms that he or she will not (i) exercise any of his or her remaining +Copyright and Related Rights in the Work or (ii) assert any associated claims +and causes of action with respect to the Work, in either case contrary to +Affirmer's express Statement of Purpose. + +4. Limitations and Disclaimers. + + a. No trademark or patent rights held by Affirmer are waived, abandoned, + surrendered, licensed or otherwise affected by this document. + b. Affirmer offers the Work as-is and makes no representations or warranties + of any kind concerning the Work, express, implied, statutory or + otherwise, including without limitation warranties of title, + merchantability, fitness for a particular purpose, non infringement, or + the absence of latent or other defects, accuracy, or the present or + absence of errors, whether or not discoverable, all to the greatest + extent permissible under applicable law. + c. Affirmer disclaims responsibility for clearing rights of other persons + that may apply to the Work or any use thereof, including without + limitation any person's Copyright and Related Rights in the Work. + Further, Affirmer disclaims responsibility for obtaining any necessary + consents, permissions or other rights required for any use of the Work. + d. Affirmer understands and acknowledges that Creative Commons is not a + party to this document and has no duty or obligation with respect to + this CC0 or use of the Work. + +For more information, please see https://creativecommons.org/publicdomain/zero/1.0/. diff --git a/README.md b/README.md index 97fc65e..257440f 100644 --- a/README.md +++ b/README.md @@ -1,303 +1,193 @@ -# 🎯 Awesome Agent Skills [![Awesome](https://awesome.re/badge.svg)](https://awesome.re) +# Awesome Agent Skills [![Awesome](https://awesome.re/badge.svg)](https://awesome.re) + +Curated, high-quality resources for building and using Agent Skills across modern AI agent platforms.
- Awesome Agent Skills - The definitive resource for Agent Skills + + Agent Skills +
-
+## Contents +- [What Are Agent Skills](#what-are-agent-skills) +- [Start Here](#start-here) +- [Phase 1: Learn the Fundamentals](#phase-1-learn-the-fundamentals) +- [Phase 2: Use Existing Skills](#phase-2-use-existing-skills) +- [Phase 3: Build and Integrate](#phase-3-build-and-integrate) +- [Phase 4: Benchmarks and Research](#phase-4-benchmarks-and-research) +- [Frequently Asked Questions](#frequently-asked-questions) -## 💡 What Are Agent Skills? +## What Are Agent Skills -**Agent Skills** mark a fundamental shift from monolithic AI systems to composable agent architectures. Rather than embedding capabilities through expensive fine-tuning or bloating context windows with static documentation, skills provide **modular, on-demand knowledge injection** through standardized `SKILL.md` packages. +Agent Skills are modular, standardized `SKILL.md` packages that give agents on-demand capabilities via progressive disclosure: lightweight metadata can load early, full instructions load only when relevant, and supporting resources are accessed when needed. -### The Architecture +Key topics: `agent-skills` · `ai-agents` · `skill-md` · `progressive-disclosure` · `context-management` · `agent-architecture` · `mcp`. -**Progressive Disclosure at Scale**: Skills leverage a three-tier context management strategy: -1. **Discovery** (~50 tokens): Lightweight metadata (name + description) loads at startup -2. **Activation** (~2-5K tokens): Full `SKILL.md` instructions load when task-relevant -3. **Execution**: Referenced scripts and resources accessed dynamically +New to Agent Skills: [agentskills.io](https://agentskills.io). -This architecture delivers: -- 🧠 **Infinite capability scaling** - No context window trade-offs -- ⚡ **Sub-second skill loading** - Zero startup latency penalty -- 🔄 **Cross-platform portability** - Write once, deploy to any compatible agent -- 📦 **Frictionless distribution** - Share via Git, install like packages +## Start Here -### The Revolution +If you are new to Agent Skills, start with these quick primers. -**From specialized agents to universal platforms**: Instead of building separate coding agents, research agents, and analysis agents, the industry is converging on **general-purpose agents with skill libraries**. One agent. Unlimited specializations. Dynamic composition. +- [What are skills](https://agentskills.io/what-are-skills) - Guide: Beginner-friendly introduction to Agent Skills. +- [Using skills in Claude](https://support.claude.com/en/articles/12512180-using-skills-in-claude) - Guide: Quick start for enabling skills in Claude. +- [Agent Skills Review](https://www.youtube.com/watch?v=tGpeWM7gOlA) - Video: Short feature highlight. -> "We used to think agents in different domains will look very different. The agent underneath is actually more universal than we thought." -> — *Barry Zhang, Anthropic Research* +## Phase 1: Learn the Fundamentals -**The open standard advantage**: Major platforms (OpenAI, GitHub, Microsoft, Cursor) have adopted the Agent Skills specification, creating **network effects**: every skill you create works across the entire ecosystem. This is the npm moment for AI agents. +Concepts, comparisons, and explainers to build a solid mental model. -### 🏷️ Key Topics +### Key Concepts and Articles -`agent-skills` · `ai-agents` · `claude-code` · `gemini-cli` ·`codex` · `skill-md` · `progressive-disclosure` · `context-management` · `ai-development` · `mcp` · `agent-architecture` · `agentic-ai` · `automation` · `productivity` +- [Equipping agents for the real world with Agent Skills](https://anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills) - Article: Original announcement from Anthropic Engineering. +- [Agent Skills Explained: Why This Changes Everything](https://www.youtube.com/watch?v=Ihoxov5x66k) - Video: Why skills matter for agent development. +- [Claude Skills vs MCP: Complete Guide](https://dev.to/jimquote/claude-skills-vs-mcp-complete-guide-to-token-efficient-ai-agent-architecture-4mkf) - Article: Comparison of Agent Skills and Model Context Protocol. +- [The Great AI Agent Configuration Confusion](https://medium.com/@satinath.mondal/the-great-ai-agent-configuration-confusion-agents-md-skill-md-and-whats-next-12345) - Article: Overview of SKILL.md, AGENTS.md, and related standards. +- [Using skills with Deep Agents](https://blog.langchain.com/using-skills-with-deep-agents/) - Article: How frameworks apply the skills pattern. -👉 **New to Agent Skills?** Start here: [agentskills.io](https://agentskills.io) +### Video Introductions -## 📚 What's Inside? +- [Don't Build Agents, Build Skills Instead](https://www.youtube.com/watch?v=CEvIs9y1uog) - Video: Anthropic talk on skills as a scalable abstraction. +- [Claude Agent Skills Tutorial and Demo](https://www.youtube.com/watch?v=mxZqEduwyFk) - Video: Intro tutorial and hands-on demo. +- [Claude's new Agent Skills](https://www.youtube.com/watch?v=VRzkafNIdgI) - Video: One-minute overview. +- [Claude Code Skills built me an AI Agent Team](https://www.youtube.com/watch?v=OdtGN27LchE) - Video: Extended beginner guide. -- [Start Here](#-start-here) - Introduction to the concepts -- [Phase 1: Learn the Fundamentals](#-phase-1-learn-the-fundamentals) - Articles and videos to get started -- [Phase 2: Use Existing Skills](#%EF%B8%8F-phase-2-use-existing-skills) - Platforms and ready-to-use libraries -- [Phase 3: Build & Integrate](#%EF%B8%8F-phase-3-build--integrate) - Create your own skills and tools -- [Phase 4: Benchmarks & Research](#-phase-4-benchmarks--research) - Advanced engineering and theory -- [Frequently Asked Questions](#-frequently-asked-questions) - Common questions answered -- [Community & Contributing](#-community--contributing) - Join the movement +## Phase 2: Use Existing Skills -## 🏁 Start Here +Platforms that support skills today, plus ready-to-use skill catalogs. -**New to Agent Skills?** These resources will take you from zero to understanding the core concepts in minutes. +### Supported Platforms and IDEs -### 🎓 Getting Started +- [Claude Code](https://claude.ai/code) - Platform: Anthropic's coding tool with skills support ([docs](https://code.claude.com/docs/en/skills)). +- [OpenAI Codex](https://developers.openai.com/codex/skills/) - Platform: OpenAI's CLI agent with skills support ([docs](https://developers.openai.com/codex/skills/)). +- [Gemini CLI](https://geminicli.com) - Gemini in the terminal with skills support ([docs](https://geminicli.com/docs/cli/skills/)). +- [Cursor](https://cursor.com/) - AI-powered editor with native skills integration ([docs](https://cursor.com/docs/context/skills)). +- [VS Code](https://code.visualstudio.com/) - Editor with Agent Skills support ([docs](https://code.visualstudio.com/docs/copilot/customization/agent-skills)). +- [GitHub Copilot](https://github.com/features/copilot) - Coding assistant with Agent Skills support ([docs](https://docs.github.com/copilot/concepts/agents/about-agent-skills)). +- [OpenCode](https://opencode.ai/) - AI development tool with built-in Agent Skills support ([docs](https://opencode.ai/docs/skills/)). +- [Amp](https://ampcode.com/) - AI coding assistant with Agent Skills support ([docs](https://ampcode.com/manual#agent-skills)). +- [Goose](https://block.github.io/goose/) - Open source agent framework with extensions support ([docs](https://block.github.io/goose/extensions)). +- [Letta](https://www.letta.com/) - Stateful LLM agents with memory ([docs](https://docs.letta.com/letta-code)). +- [Roo Code](https://roocode.com/) - VS Code extension and cloud agents with skills integration ([docs](https://docs.roocode.com/features/skills)). -- [What are skills?](https://agentskills.io/what-are-skills) - Perfect introduction for beginners -- [Using skills in Claude](https://support.claude.com/en/articles/12512180-using-skills-in-claude) - Quick start guide -- [Agent Skills Review](https://www.youtube.com/watch?v=tGpeWM7gOlA) - 25-second feature highlight +### Ready-to-Use Skill Libraries -## 🎓 Phase 1: Learn the Fundamentals +#### Top Picks -**Understand the "Why" and "How".** Before diving into code, grasp the architectural shift and best practices. +- [Anthropic skills](https://github.com/anthropics/skills) - Library: Official skills catalog from Anthropic. +- [OpenAI skills](https://github.com/openai/skills) - Library: Official skills catalog from OpenAI. +- [Hugging Face skills](https://github.com/huggingface/skills) - Library: Community skills catalog with broad compatibility. +- [karanb192/awesome-claude-skills](https://github.com/karanb192/awesome-claude-skills) - Directory: Curated list of Claude skills. +- [shajith003/awesome-claude-skills](https://github.com/shajith003/awesome-claude-skills) - Directory: Skill collection for specialized capabilities. -### 📰 Key Concepts & Articles +#### More Collections -- [Equipping agents for the real world with Agent Skills](https://anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills) - The original announcement from Anthropic Engineering -- [Agent Skills Explained: Why This Changes Everything](https://www.youtube.com/watch?v=Ihoxov5x66k) - 17-minute analysis of why Skills matter for AI development -- [Claude Skills vs MCP: Complete Guide](https://dev.to/jimquote/claude-skills-vs-mcp-complete-guide-to-token-efficient-ai-agent-architecture-4mkf) - Comparison of Skills and Model Context Protocol architectures -- [The Great AI Agent Configuration Confusion](https://medium.com/@satinath.mondal/the-great-ai-agent-configuration-confusion-agents-md-skill-md-and-whats-next-12345) - Understanding configuration standards (SKILL.md, AGENTS.md) -- [Using skills with Deep Agents](https://blog.langchain.com/using-skills-with-deep-agents/) - How generalist agents use skills for efficiency +- [GuDaStudio/skills](https://github.com/GuDaStudio/skills) - Multi-agent collaboration skills. +- [DougTrajano/pydantic-ai-skills](https://github.com/DougTrajano/pydantic-ai-skills) - Pydantic AI integration skills. +- [OmidZamani/dspy-skills](https://github.com/OmidZamani/dspy-skills) - Skills for DSPy-based workflows. +- [ponderous-dustiness314/awesome-claude-skills](https://github.com/ponderous-dustiness314/awesome-claude-skills) - Document editing, data analysis, and project management skills. +- [hikanner/agent-skills](https://github.com/hikanner/agent-skills) - Curated Agent Skills collection. +- [gradion-ai/freeact-skills](https://github.com/gradion-ai/freeact-skills) - Freeact skill library. -### 🎬 Video Introductions -- [Don’t Build Agents, Build Skills Instead (Anthropic)](https://www.youtube.com/watch?v=CEvIs9y1uog) - Why “skills” are the scalable abstraction for agent capability -- [A better way to build agents? Claude Agent Skills Tutorial + Demo](https://www.youtube.com/watch?v=mxZqEduwyFk) - 11-minute intro: Is Agent Skills the next big standard after MCP? -- [Claude's new 'Agent Skills'](https://www.youtube.com/watch?v=VRzkafNIdgI) - 1-minute quick overview - installable "mini-brains" -- [Claude Code Skills just Built me an AI Agent Team (2026 Guide)](https://www.youtube.com/watch?v=OdtGN27LchE) - 27-minute beginner guide to building general AI agents +### Skill Marketplaces -## 🛠️ Phase 2: Use Existing Skills +- [SkillsMP](https://skillsmp.com/) - Marketplace for discovering and sharing Agent Skills. +- [Skillstore](https://skillstore.io/) - Curated marketplace for Agent Skills. +- [SkillsDirectory](https://www.skillsdirectory.org/) - Directory of popular Agent Skills. -**Experience the power.** Don't build from scratch—leverage the ecosystem of platforms and pre-built libraries. +## Phase 3: Build and Integrate -### 🎨 Supported Platforms & IDEs +Guides and tools for authoring, validating, and distributing skills. -**Where can you use Agent Skills today?** +### How to Build Skills -- [Claude Code](https://claude.ai/code) - Anthropic's coding tool with first-class skills support ([docs](https://code.claude.com/docs/en/skills)) -- [OpenAI Codex](https://developers.openai.com/codex/skills/) - OpenAI's CLI agent with Agent Skills support ([docs](https://developers.openai.com/codex/skills/)) -- [Gemini CLI](https://geminicli.com) - Google's solution for bringing Gemini into the terminal ([docs](https://geminicli.com/docs/cli/skills/)) -- [Cursor](https://cursor.com/) - AI-powered code editor with native skills integration ([docs](https://cursor.com/docs/context/skills)) -- [VS Code](https://code.visualstudio.com/) - Microsoft's editor with Agent Skills in Insiders build ([docs](https://code.visualstudio.com/docs/copilot/customization/agent-skills)) -- [GitHub Copilot](https://github.com/features/copilot) - GitHub's AI coding assistant with Agent Skills support ([docs](https://docs.github.com/copilot/concepts/agents/about-agent-skills)) -- [OpenCode](https://opencode.ai/) - AI development tool with built-in Agent Skills support ([docs](https://opencode.ai/docs/skills/)) -- [Amp](https://ampcode.com/) - Next-gen AI coding assistant ([docs](https://ampcode.com/manual#agent-skills)) -- [Goose](https://block.github.io/goose/) - Open source AI agent framework ([docs](https://block.github.io/goose/extensions)) -- [Letta](https://www.letta.com/) - Build stateful LLM agents with memory ([docs](https://docs.letta.com/letta-code)) -- [Roo Code](https://roocode.com/) - AI dev team through a VSCode extension and cloud agents ([docs](https://docs.roocode.com/features/skills)) +- [How to create custom skills](https://support.claude.com/en/articles/12512198-creating-custom-skills) - Guide: Step-by-step instructions for authoring skills. +- [Skills API Quickstart](https://docs.claude.com/en/api/skills-guide#creating-a-skill) - Docs: API reference for implementing skills. +- [How I Built Agent Skills for Claude Code](https://dev.to/nunc/how-i-built-agent-skills-for-claude-code-oj4) - Tutorial: Practical walkthrough for building custom skills. +- [Claude Agent Skills Tutorial](https://www.youtube.com/watch?v=fOxC44g8vig) - Video: End-to-end walkthrough with examples. -### 📦 Ready-to-Use Skill Libraries +### Developer Tools -**Tap into battle-tested capabilities.** +- [LangChain Multi-Agent Skills](https://docs.langchain.com/oss/python/langchain/multi-agent/skills) - Docs: Implementing skills in LangChain. +- [SkillCheck](https://github.com/agentigy/skillcheck) - Tool: Scanner for common risks in skill packages. +- [OpenSkills](https://github.com/numman-ali/openskills) - Tool: Universal loader for integrating skills with many agents. +- [LangChain Deep Agents](https://github.com/langchain-ai/deepagents) - Framework: Agent harness with a skills-oriented workflow. +- [IntentKit](https://github.com/crestalnetwork/intentkit) - Framework: Intent-driven agent building. +- [Agentica](https://github.com/wrtnlabs/agentica) - Framework: TypeScript function-calling utilities for agents. -#### 🔥 Top Picks -- Official collection from Anthropic (document editing, data analysis, etc.) -- Official collection from OpenAI (Codex skills catalog) -- Official collection from Hugging Face (compatible with Claude, Codex, Gemini) -- 50+ verified skills for Claude Code and Claude.ai -- Skills for specialized capabilities +### Reference Implementations -#### 📂 More Collections -- Multi-agent collaboration skills -- Pydantic AI integration -- Skills for DSPy framework -- Document editing, data analysis, and project management -- Curated Claude Agent Skills collection -- Freeact agent library skills +#### Development and Programming -### 🏪 Skill Marketplaces -- [SkillsMP](https://skillsmp.com/) - Agent Skills marketplace for Claude Code/Codex (discover, install, and share skills) -- [Skillstore](https://skillstore.io/) - Curated, security-audited Agent Skills marketplace for Claude Code and Codex -- [SkillsDirectory](https://www.skillsdirectory.org/) - Directory for finding popular Agent Skills +- [kylehughes/the-unofficial-swift-concurrency-migration-skill](https://github.com/kylehughes/the-unofficial-swift-concurrency-migration-skill) - Skill-style guide for Swift concurrency migrations. +- [gapmiss/obsidian-plugin-skill](https://github.com/gapmiss/obsidian-plugin-skill) - Skill package for Obsidian plugin development. +- [frmoretto/stream-coding](https://github.com/frmoretto/stream-coding) - Stream-coding methodology reference. -## 🏗️ Phase 3: Build & Integrate +#### Integration and Automation -**Create your own.** Learn to author `SKILL.md` files and integrate them into your workflows. +- [SawyerHood/dev-browser](https://github.com/SawyerHood/dev-browser) - Browser capability for agents. +- [gotalab/skillport](https://github.com/gotalab/skillport) - CLI and MCP-based skill distribution. +- [gmickel/sheets-cli](https://github.com/gmickel/sheets-cli) - Google Sheets automation via CLI. +- [fabioc-aloha/spotify-skill](https://github.com/fabioc-aloha/spotify-skill) - Spotify API integration skill. -### 📝 How to Build Skills +## Phase 4: Benchmarks and Research -- [How to create custom skills](https://support.claude.com/en/articles/12512198-creating-custom-skills) - Step-by-step authoring guide -- [Skills API Quickstart](https://docs.claude.com/en/api/skills-guide#creating-a-skill) - Technical documentation -- [How I Built Agent Skills for Claude Code](https://dev.to/nunc/how-i-built-agent-skills-for-claude-code-oj4) - 8-minute tutorial on building custom Agent Skills from scratch -- [Claude Agent Skills Tutorial](https://www.youtube.com/watch?v=fOxC44g8vig) - Comprehensive implementation tutorial with SKILL.md format +Evaluation frameworks and deeper technical reading. -### 🔧 Developer Tools +### Benchmarks and Evaluation -**Frameworks and utilities for building and loading skills.** +- [benchflow-ai/SkillsBench](https://github.com/benchflow-ai/SkillsBench) - Benchmark: Measuring skills performance on real workflows. -- [LangChain Multi-Agent Skills](https://docs.langchain.com/oss/python/langchain/multi-agent/skills) - Official documentation on implementing the skills pattern in LangChain -- [SkillCheck](https://github.com/agentigy/skillcheck) - Security scanner for SKILL.md files - detect vulnerabilities before production -- [OpenSkills](https://github.com/numman-ali/openskills) - Universal skills loader for any AI agent -- [LangChain Deep Agents](https://github.com/langchain-ai/deepagents) - Open source agent harness with skills support -- [IntentKit](https://github.com/crestalnetwork/intentkit) - Framework for intent-driven AI agents -- [Agentica](https://github.com/wrtnlabs/agentica) - TypeScript AI function calling framework +### Advanced Engineering -### 💡 Reference Implementations (By Category) +- [Claude Agent Skills: A First Principles Deep Dive](https://leehanchung.github.io/blogs/2025/10/26/claude-skills-deep-dive/) - Article: Detailed architecture analysis. +- [I finally CRACKED Claude Agent Skills](https://www.youtube.com/watch?v=kFpLzCVLA20) - Video: Comparison of Skills, MCP, and subagents. +- [Claude Agent Skills](https://www.youtube.com/watch?v=9XaprFRNTlc) - Video: One-hour deep dive into domain-specific usage. +- [muratcankoylan/Agent-Skills-for-Context-Engineering](https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering) - Repository: Context-engineering reference materials. +- [jakedahn/pomodoro](https://github.com/jakedahn/pomodoro) - Repository: System skill pattern reference. +- [yzfly/Mind-Cloning-Engineering](https://github.com/yzfly/Mind-Cloning-Engineering) - Repository: Techniques for building reusable skill knowledge. -**Learn from examples.** +### Academic Papers -#### Development & Programming -- Swift Concurrency Migration guide -- Obsidian.md plugin development -- Stream Coding methodology +- [Agent Skills Enable a New Class of Realistic and Trivially Simple Prompt Injections](https://arxiv.org/abs/2510.26328) (2025) - Paper: Security analysis of skill-file prompt injection risks. +- [A survey of agent interoperability protocols](https://arxiv.org/abs/2505.02279) (2025) - Paper: Survey of MCP, Agent Cards, and related protocols. +- [Reinforcement Learning for Self-Improving Agent with Skill Library](https://arxiv.org/abs/2512.17102) (2024) - Paper: Maintaining and improving skill libraries. +- [PolySkill: Learning Generalizable Skills Through Polymorphic Abstraction](https://arxiv.org/abs/2510.15863) (2024) - Paper: Learning transferable skills via abstraction. -#### Integration & Automation -- Web browser capability for agents -- Skills for any agent via CLI or MCP -- Google Sheets CLI automation -- Spotify API integration +## Frequently Asked Questions -## 🔬 Phase 4: Benchmarks & Research +### What are Agent Skills -**Deepen your expertise.** Explore evaluations, security implications, and academic research. +Agent Skills are modular `SKILL.md` packages that provide on-demand capabilities without loading all knowledge up front. -### 📏 Benchmarks & Evaluation -- SkillsBench - Benchmark/evaluation framework for measuring Agent Skills performance on real workflows +### How do Agent Skills differ from fine-tuning -### 🧠 Advanced Engineering +Fine-tuning changes model weights, while skills provide runtime knowledge and workflows that you can update instantly. -- [Claude Agent Skills: A First Principles Deep Dive](https://leehanchung.github.io/blogs/2025/10/26/claude-skills-deep-dive/) - Comprehensive analysis of the internal architecture -- [I finally CRACKED Claude Agent Skills](https://www.youtube.com/watch?v=kFpLzCVLA20) - Engineering breakdown: Skills vs MCP vs Subagents -- [Claude Agent Skills](https://www.youtube.com/watch?v=9XaprFRNTlc) - 1-hour deep dive into domain-specific AI expertise -- Context engineering techniques -- System Skill Pattern (skills that remember & improve) -- Mind cloning with LLM skills +### What is the difference between Agent Skills and MCP -### 🛡️ Academic Papers +Agent Skills focus on workflows and capabilities, while MCP focuses on secure, structured data and tool access. -- [Agent Skills Enable a New Class of Realistic and Trivially Simple Prompt Injections](https://arxiv.org/abs/2510.26328) (2025) - Security analysis showing vulnerabilities in Agent Skills markdown files -- [A survey of agent interoperability protocols](https://arxiv.org/abs/2505.02279) (2025) - Comprehensive look at MCP, Agent Cards, and more -- [Reinforcement Learning for Self-Improving Agent with Skill Library](https://arxiv.org/abs/2512.17102) (2024) - Framework for agents to build and maintain skill libraries -- [PolySkill: Learning Generalizable Skills Through Polymorphic Abstraction](https://arxiv.org/abs/2510.15863) (2024) - Learning generalizable skills through polymorphic abstraction +### How do I create my first Agent Skill -## ❓ Frequently Asked Questions +See the [How to Build Skills](#how-to-build-skills) section for a step-by-step authoring guide. -### What are Agent Skills? +### Which AI platforms support Agent Skills -Agent Skills are modular, standardized `SKILL.md` packages that provide AI agents with on-demand capabilities. Instead of fine-tuning models or bloating context windows, skills enable **progressive disclosure**: lightweight metadata loads at startup (~50 tokens), full instructions activate when needed (~2-5K tokens), and resources load dynamically during execution. +Support varies by platform, but major tools include Claude (Claude.ai and Claude Code), OpenAI Codex, GitHub Copilot, Cursor, VS Code, and more. -### How do Agent Skills differ from fine-tuning? +### Can I use Agent Skills with ChatGPT or other LLMs -Fine-tuning modifies model weights permanently (expensive, inflexible), while Agent Skills provide **runtime knowledge injection** that's instantly updatable, shareable across platforms, and requires zero retraining. Update a skill once, and every agent using it benefits immediately. +If a platform does not support the format natively, you can often integrate skills via a loader or by adapting the `SKILL.md` instructions into that platform's prompt workflow. -### What's the difference between Agent Skills and MCP (Model Context Protocol)? +### Are Agent Skills secure -**Agent Skills** focus on **capabilities and workflows** (how to do X), while **MCP** focuses on **data access** (connecting to APIs, databases). They're complementary: use MCP to connect to external data sources, use Skills to teach agents how to process that data. Many developers use both together. +Treat skills like code: review them before using, avoid installing from untrusted sources, and prefer audited skill libraries. -### How do I create my first Agent Skill? +### How do I share Agent Skills with my team -1. Create a `SKILL.md` file with YAML frontmatter (name, description) -2. Write clear instructions in markdown (what, when, how) -3. Add optional `scripts/` folder for code references -4. Place in `.github/skills/` or `.claude/skills/` directory -5. Test with compatible platforms (Claude Code, Cursor, GitHub Copilot) +The most common approach is to version skills in Git (in a shared repo) and let your supported tools discover them from a standard directory. -👉 Full guide: [How to create custom skills](https://support.claude.com/en/articles/12512198-creating-custom-skills) +## Contributing -### Which AI platforms support Agent Skills? - -Major platforms with native support include: **Claude** (Claude.ai, Claude Code), **OpenAI** (Codex CLI), **GitHub Copilot**, **Cursor**, **VS Code Insiders**, **OpenCode**, **Amp**, **Letta**, and **Goose**. The ecosystem is rapidly expanding as more platforms adopt the open standard. - -### Can I use Agent Skills with ChatGPT or other LLMs? - -While ChatGPT doesn't natively support the Agent Skills format, you can use universal loaders like [openskills](https://github.com/numman-ali/openskills) to integrate skills with any LLM platform. The skills specification is open and platform-agnostic by design. - -### Are Agent Skills secure? - -Agent Skills introduce new security considerations. Since skills can execute code and access files, always **review skills before installation** from untrusted sources. Use tools like [skillcheck](https://github.com/agentigy/skillcheck) to scan for vulnerabilities. Research shows potential for prompt injection attacks ([arxiv.org/abs/2510.26328](https://arxiv.org/abs/2510.26328)), so treat skills like code packages—audit before trusting. - -### How do I share Agent Skills with my team? - -**Option 1**: Commit to your repository's `.github/skills/` or `.claude/skills/` directory—agents automatically discover them. -**Option 2**: Publish to GitHub as a standalone repo (e.g., `my-org/data-analysis-skill`)—others can reference or clone. -**Option 3**: Use [skillport](https://github.com/gotalab/skillport) or [agent-skills-mcp](https://github.com/DiscreteTom/agent-skills-mcp) for cross-platform distribution. - -### What makes a good Agent Skill? - -**Best practices:** -- **Single responsibility** - One clear capability per skill -- **Progressive detail** - Brief metadata, detailed instructions, extensive resources -- **Context-aware** - Include when/why to use (not just how) -- **Testable** - Provide example inputs/outputs -- **Discoverable** - Descriptive names and rich descriptions - -### When should I use Agent Skills vs custom tools/functions? - -Use **Agent Skills** when: -- Capabilities need to work across multiple platforms -- Instructions are complex multi-step workflows -- You want version control and collaborative editing -- Knowledge needs frequent updates - -Use **custom tools/functions** when: -- Platform-specific integration required (e.g., API-specific features) -- Real-time data access needed -- Computational logic can't be expressed as instructions - -### Can Agent Skills call other Agent Skills? - -Yes! Skills can reference other skills in their instructions, enabling **skill composition**. This allows building complex capabilities from simple, reusable building blocks. Some platforms also support **skill libraries** where skills automatically discover and leverage related capabilities. - -### How do Agent Skills impact token usage? - -**Dramatically reduce it.** Traditional approaches load all documentation upfront (10K-100K tokens), while skills load only what's needed (50 tokens metadata + 2-5K when activated). For agents managing 50+ capabilities, this means **90%+ token reduction** during idle time. - -### Are Agent Skills just for coding tasks? - -No—Agent Skills work for any domain: **data analysis**, **content creation**, **project management**, **research**, **security operations**, **automation**, and more. The format is domain-agnostic. Think of them as "packages for knowledge" rather than "packages for code." - -### Where can I find production-ready Agent Skills? - -Start with: -- [anthropics/skills](https://github.com/anthropics/skills) - Official Anthropic collection -- [karanb192/awesome-claude-skills](https://github.com/karanb192/awesome-claude-skills) - 50+ verified skills -- This repository's [Ready-to-Use Skills](#-ready-to-use-skills) section - -### How often should I update my Agent Skills? - -Update skills whenever: -- Platform APIs or tools change -- Better approaches emerge -- Team workflows evolve -- User feedback reveals gaps - -**Pro tip**: Add version numbers and changelogs to your skills for better tracking. - -## 💬 Join the Community - -**You're not alone in this.** Join thousands of developers pioneering the future of AI agents - share breakthroughs, debug challenges, and shape the standards. - -- 💭 [Agent Skills GitHub Discussions](https://github.com/agentskills/agentskills/discussions) - Official specification discussions -- 🐛 [Anthropic Skills GitHub Issues](https://github.com/anthropics/skills/issues) - Report issues and request features -- ⭐ **Star this repo** to stay updated with the latest resources! - -## 🤝 Contributing - -**Shape the ecosystem.** This is a community-driven resource - your contributions help thousands of developers discover and build better agent systems. - -- Add new skills, tools, or articles -- Improve documentation -- Fix broken links -- Suggest new sections - -Please read our [contribution guidelines](CONTRIBUTING.md) before submitting. - ---- - -
- -**Found this useful? Give it a ⭐ and share it with others!** - -*Built with ❤️ by the Agent Skills community* - -
+Please read our [contribution guidelines](contributing.md) before submitting changes. diff --git a/contributing.md b/contributing.md new file mode 100644 index 0000000..eb08493 --- /dev/null +++ b/contributing.md @@ -0,0 +1,25 @@ +# Contributing to Awesome Agent Skills + +Thank you for your interest in contributing! We welcome new resources that help the community understand and use Agent Skills. + +## How to Contribute + +1. Fork the repository. +2. Create a new branch: `git checkout -b add-resource`. +3. Add your resource to the appropriate section in `README.md`. + - Please follow the existing format: `[Title](URL) - Description.` + - Keep descriptions concise. + - Ensure the link is valid. +4. Commit your changes: `git commit -m "Add [Resource Name]"`. +5. Push to your branch: `git push origin add-resource`. +6. Submit a Pull Request. + +## Guidelines + +- **Relevance**: Ensure the resource is directly related to "Agent Skills" (the `SKILL.md` format) or closely related concepts like progressive disclosure for agents. +- **Quality**: Link to high-quality, up-to-date content. +- **No Duplicates**: Check if the resource is already listed. + +## License + +By contributing, you agree that your contributions will be licensed under the same license as the project.