Agent skills

Agent Skills is an open specification for packaging reusable, task-scoped bundles of context that AI coding agents can load on demand. Skills support context engineering.

A skill is simply a directory containing a SKILL.md file — YAML front-matter followed by Markdown instructions. The front-matter’s description field is the primary activation mechanism. At startup, the agent reads only each skill’s name and description. The full body is loaded only when the description matches the current task. This on-demand, just-in-time loading keeps the context window focused on what is currently relevant.

In this sense a skill is the unit of progressive disclosure in a harness. Only the name and description sit in context by default; the full instructions arrive only when triggered. This is what distinguishes a skill from AGENTS.md, which is loaded into every session regardless of task. See agent memory for the wider memory system this fits into.

Skills can also contain optional scripts/, references/, and assets/ subdirectories for executable code, supplementary documentation, and static resources.

The canonical specification lives at https://agentskills.io/specification. The specification is supported by a growing set of coding agent harnesses. Some systems call them "rules" or "instructions".

In the language of agent harness engineering, skills are a guide — a feed-forward control that steers an agent toward the right approach before it acts, rather than a tool it executes or a sensor that checks its output.

Agent skills complement the AGENTS.md convention. AGENTS.md is the always-on project orientation, loaded into every session. Skills encapsulate the procedures, standards, and playbooks that are relevant only to certain tasks. See agent memory for how AGENTS.md and skills fit the broader memory system.

Skills registries and collections

  • Agent Skills Directory (skills.sh): The primary public registry for discoverable, installable skills. Install skills into any project via npx skillsadd <owner/repo>.
  • Anthropic Skills: A collection of skills maintained by Anthropic, optimized for use with Claude models.
  • Google Skills: A collection of domain-specific skills for working with Google products.
  • OpenAI Actions: Domain-specific tools and actions (not agent skills, specifically) for use with OpenAI’s models and agents SDK.

See also