Back to directory

Agent Skills Ultimate Guide

A guide and curated collection of Agent Skill resources, tutorials, and practical examples, explaining the open Agent Skills specification and how to install and structure a skill.

Agent Skills Ultimate Guide

This project collects and shares Agent Skill resources, tutorials, and practical examples, meant to help more people take the first step toward building a personalized agent. A skill packages a specific business process or domain knowledge so a user does not need to repeat background context in every conversation.

It documents the standard skill folder structure: a required SKILL.md file whose YAML frontmatter names the skill and describes what it does and when to use it, plus optional scripts, references, and assets folders. An agent loads only a skill's name and description at startup, loads its full SKILL.md once a task matches, and reads its scripts, references, or assets only as needed.

The underlying Agent Skills specification, started by Anthropic and maintained with the community, is supported by hosts including Claude Code, ChatGPT, Codex, GitHub Copilot, Cursor, Gemini CLI, VS Code, OpenCode, Kiro, and JetBrains Junie. Skills are installed by placing their folder in a project-level or user-level skills directory, or through a third-party command-line tool that supports many agent hosts.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.