Back to directory

skills

A command-line tool for installing and using AI agent skills from a Git repository, a URL or an archive, compatible with Claude Code, Codex, Cursor, OpenCode and dozens of other coding agents.

skills

skills add installs a skill's files into a project's local skills directory, or a user's global one, either as a copy or a symlink, targeted at a chosen agent; skills use instead generates the equivalent prompt for a single skill without installing it, and can pipe that prompt straight into a supported coding agent to start it interactively.

It resolves several source formats, including GitHub shorthand references and private repositories, trying the user's existing Git credentials first, then the GitHub CLI's stored authentication, then SSH, and falling back to an authenticated clone if anonymous or token-based GitHub API lookups fail; a GITHUB_TOKEN or GH_TOKEN environment variable can also be set explicitly when needed. A skill can also be downloaded directly from a URL pointing at a SKILL.md file or a .zip, .tar, .tar.gz or .tgz archive, with configurable size limits on the download, its extracted contents, and the number of files an archive may contain.

It is written in TypeScript and licensed under the MIT license.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.