Seonglae ChoInstead of full description, give an index of capabilities
Agent Skills is an open standard in the form of a bundle of procedures, scripts, and resources (folder) that agents can call upon when needed.
A very simple "capability pack" composed of Markdown + scripts.
More token-efficient compared to MCP (only YAML metadata at the beginning is loaded into the session). Skills are simple, flexible, separated by task units, and extensible like using CLI tools.
- AI-research-SKILLsOrchestra-Research • Updated 2026 Feb 3 14:26
npx @orchestra-research/ai-research-skills
- find-skillagent-skillsvercel-labs • Updated 2026 Jan 20 0:54gemini-skillsgoogle-gemini • Updated 2026 May 13 12:24skillsgoogle • Updated 2026 May 13 13:33skillsmicrosoft • Updated 2026 May 13 12:22
- skillselevenlabs • Updated 2026 May 13 12:51
npx skills add elevenlabs/skills
Agent Skills Techniques
A benchmark created to measure how much Agent Skills (procedural task knowledge packages) actually improve performance.
Human-created Skills are effective with an average +16.2%p performance increase. Model-generated Skills are nearly ineffective. Models cannot generate Procedural Knowledge at a level they can effectively consume.
arxiv.org
https://arxiv.org/pdf/2602.12670
Agent Skills
Overview - Agent Skills
A simple, open format for giving agents new capabilities and expertise.
https://agentskills.io/home
Claude Skills are awesome, maybe a bigger deal than MCP
Anthropic this morning introduced Claude Skills, a new pattern for making new abilities available to their models: Claude can now use Skills to improve how it performs specific tasks. Skills …
https://simonwillison.net/2025/Oct/16/claude-skills/
React skill from vercel
Introducing: React Best Practices - Vercel
We've encapsulated 10+ years of React and Next.js optimization knowledge into react-best-practices, a structured repository optimized for AI agents and LLMs.
https://vercel.com/blog/introducing-react-best-practices
Concordance AI Control
Token injection likeSteering Vector https://github.com/concordance-co/quote
Token Injection as a Steering Mechanism for Large Language Models
Lightweight steering of LLMs through token injection at inference time
https://www.concordance.co/blog/token-injection-steering-llms
