Skip to main content

​
Welcome to FastSkill

Package manager and operational toolkit for Agent AI Skills. FastSkill follows the SKILL.md skill layout and adds manifests, lockfiles, validation, search, evals, and optional operator publishing.

​
Highlights

  • Manage skill dependencies with skill-project.toml and skills.lock
  • Install from local folders, git repositories, or registries
  • Search skills across remote catalogs or local installs
  • Validate and evaluate skills before sharing
  • Diagnose your environment with fastskill doctor
  • Publish skills only when the optional registry feature is enabled

​
Quick start

​
Features

​
Projects

Use skill-project.toml to define dependencies and skills.lock to pin them.

​
Discovery

Search by meaning, or stay local when you only want installed skills.

​
Validation and evals

Use fastskill list, fastskill eval validate, and fastskill doctor to check readiness before you share skills.

​
Distribution

fastskill package builds skill artifacts. Registry publish is available only in builds that enable registry-publish.

​
How it fits together

  • Authors edit SKILL.md, install dependencies, validate, and package when ready.
  • Developers install, list, read, and search installed skills.
  • Teams keep shared manifests and locks for reproducible installs.
  • Automation uses fastskill install --lock and the same commands as local workflows.

​
Use cases

  • Individuals: manage a personal skill library and keep installs reproducible.
  • Teams: share manifests and locks across teammates.
  • Organizations: operate many skills with consistent versions and environment groups.
  • Skill authors: validate, evaluate, package, and optionally publish when ready.

​
Next steps