Agent Skills provides a standardized, open format for extending AI agents with reusable, folder-based capabilities. It documents a SKILL.md-driven structure and a three-stage loading model (Discovery, Activation, Execution).
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Agent Skills offers a lightweight, open format for extending AI agent capabilities with specialized knowledge and workflows. A skill is a folder containing a SKILL.md file with metadata and instructions, and may include optional subfolders like scripts/, references/, and assets/.
How it works
Agents load skills via progressive disclosure in three stages:
- Discovery: at startup, load only name and description of each skill.
- Activation: when a task matches a skill's description, read the full
SKILL.mdinto context. - Execution: follow the instructions, optionally executing bundled code or loading referenced files.
Full instructions load only when a task calls for them, preserving a small context footprint while keeping many skills on hand.
Getting started
- Documentation:
https://agentskills.io— Guides and tutorials - Specification:
https://agentskills.io/specification— Format details - Example Skills:
https://github.com/anthropics/skills— See what's possible - Discord:
https://discord.gg/MKPE9g8aUy— Share what you're building!
Recent releases
RELEASES (latest 0):
- none
Traction
stars: 23834 forks: 1706 open_issues: 51 language: Python license: Apache-2.0 created: 2025-12-16 last_push: 2026-08-03 topics: agent-skills
Behind the repo
The repository notes open development origins and contributions by the broader ecosystem; see LICENSE and documentation licenses for details.
Caveats
License: Apache-2.0 for code; CC-BY-4.0 for documentation. No specific age-related caveats beyond creation date provided.