Front page / Concepts / Agent skills
Agent skills
A skill is a folder with a short description and instructions, sometimes with reference files and scripts, that an assistant loads only when the description matches what you asked. Instead of re-explaining how you want a weekly update written, you install the skill once and the assistant picks it up by itself. It is how a method gets packaged, shared and installed.
Where it stands
Among the 14 concepts. Bars: GitHub stars gained this week.
- MCP and tool use+3.9k
- Agentic loops+1.3k
- Knowledge graphs (GraphRAG)+1.2k
- Retrieval (RAG)+1.2k
- Prompt engineering+1.1k
- Vibe coding+980
- Workflows, not agents+565
- Spec-driven work+16k
- Memory+7.3k
- Agent skills+5.7k
- Multi-agent orchestration+5.6k
- Computer use+4.7k
- Context engineering+3.3k
- Evals and verification+929
How to use it today
In ChatGPT or Claude, as a person doing the work, not as a developer.
- 1Install a skill from this catalog into your assistant with the Add button on its listing page (one package into one app).
- 2Start a new session and ask for the work in plain words, for example "turn these notes into a problem statement"; the assistant matches the request to the skill's description.
- 3Read the skill before relying on it: open its SKILL.md on the listing page and check that its steps are the ones you would take.
- 4When a skill is close but not right, copy it into your own folder and edit the instructions; a skill is text.
- 5Remove skills you do not use: every installed description is read at the start of a session.
What changes day to day
Your recurring asks become one-line requests. The quality of a weekly update, a spec or a readout stops depending on the day, because the method is written down and loaded every time.
Where it goes wrong
Installing dozens of skills at once means none of them is understood and the assistant has more descriptions to choose among. A skill written for a developer's workflow will not fit a product manager's without editing.
Listings that use it well
Chosen by hand after reading them, each with the reason. Install from the listing's own page.
| Listing | Kind | Why it is here | Stars |
|---|---|---|---|
| skill-creator | Agent skills | Anthropic's reference skill for creating, editing and evaluating skills, the canonical SKILL.md workflow. | 190k |
| ai-engineering-from-scratch | Agent skills | A course path on Agent Skills engineering plus skill-catalog-builder, skill-contract-reviewer, skill-invocation-router and skill-release-gate. | 78k |
| book-to-skill | Agent skills | Turns any technical book into a skill folder, the clearest demonstration of packaging knowledge for on demand loading. | 37k |
| one-skill-to-rule-them-all | Agent skills | A meta skill that watches work sessions and turns corrections and patterns into new skills and skill improvements. | 3.8k |
| skillkit | Agent skills | Installs, translates and shares skills across forty agent clients, with a skill-authoring skill for SKILL.md structure. | 1.6k |
Elsewhere
Repositories that embody the idea and are not in the catalog.
- agentskills/agentskills ↗The open specification and reference material for the SKILL.md format.
Standing
Our judgment, with the reason and the sources it rests on.
Rising since 2025-10. Anthropic published the format in October 2025 and most assistants now read the same folder layout, which is what this catalog installs. Introduced for Claude on 2025-10-16, published as an open standard at agentskills.io on 2025-12-18, and by 2026 supported by Codex, Gemini CLI, GitHub Copilot, Cursor, Goose and dozens of other clients listed on the standard's site. In 2026 the same format reached the chat apps: ChatGPT plugins and Microsoft's Copilot both carry skills. Adopted faster each month; the vendors are building for it and new tools keep appearing.
- Introducing Agent Skills2025-10-16
- Skills for organizations, partners, the ecosystem2025-12-18
- Agent Skills (Simon Willison)2025-12-19
Related: Context engineering, MCP and tool use, Workflows, not agents