What Are Agent Skills?
Learn what Agent Skills are, how they work, and why they matter. Understand the open format for extending AI agent capabilities with specialized knowledge and workflows.
How skills work
Skills use a three-step process called progressive disclosure to remain efficient:
Discovery
When an agent starts up, it loads only the name and description of each available skill. This allows the agent to know when a skill might be relevant without consuming much context (about 50 to 100 tokens per skill).
Activation
When a user's task matches a skill's description, the agent reads the full SKILL.md instructions into its context window.
Execution
The agent follows the instructions, loading additional reference files or running bundled scripts only as needed.
This approach keeps agents fast and focused. Instead of loading everything at once, skills provide context on demand.
What's inside a skill?
Every skill begins with a SKILL.md file. This Markdown file contains metadata and instructions. Skills can also include extra resources:
The SKILL.md file uses YAML frontmatter for metadata and Markdown for instructions:
Why this format?
Universal
Works across Claude, ChatGPT, Cursor, Copilot, Codex, Antigravity, and many other AI tools.
Human-readable
Anyone can open a SKILL.md file and understand what it does.
Portable
Skills are just files; easy to share, version, and edit.
Extensible
From simple text instructions to bundled scripts, templates, and data.
What can skills do?
Skills can package any type of AI ability:
Skills on Skly
On Skly, skills follow the Agent Skills format. When you buy a skill, you get a bundle you can use with any compatible AI tool. When you sell a skill, you are packaging your expertise into a format that works everywhere, giving buyers immediate value across their entire AI toolkit.