The card catalog for AI agent capabilities

What it does This is an awesome-list that tracks the Agent Skills ecosystem: modular SKILL.md packages that give AI agents on-demand capabilities through progressive disclosure. It rounds up official docs, tutorials, video guides, skill marketplaces, and academic papers so developers can find platforms, libraries, and research without hunting through scattered sources.

The interesting bit The list captures how fast a de facto standard can solidify. In a few months, Anthropic, OpenAI, Google, Microsoft, Cursor, and VS Code have all aligned on the same SKILL.md pattern—complete with official skill libraries, security benchmarks, and even dedicated marketplaces like SkillsMP.

Key highlights

• Maps 13 supported platforms including Claude Code, OpenAI Codex, Gemini CLI, Cursor, and GitHub Copilot
• Catalogs official skill libraries from Anthropic, OpenAI, Microsoft, Google, and Hugging Face
• Tracks marketplaces and directories such as agentskill.sh (44k+ skills) and SkillsMP
• Includes security research on skill-file prompt injection and 2026 academic papers on skill graphs and lifelong discovery
• Covers the full lifecycle from beginner primers to build guides and evaluation frameworks

Caveats

• It is a curated link directory, not a framework or executable tool
• The README is structured as a table of contents with minimal commentary of its own

Verdict Worth bookmarking if you are building or consuming agent capabilities and want to see where the SKILL.md standard is heading. Skip it if you are looking for a single tool to install; this is purely a reading list.