Back to directory

toolkit-ai

A package manager for AI coding assistants that installs skills, subagents and MCP servers from any GitHub or Bitbucket repository into Claude Code, Codex, Amp, GitHub Copilot and Cursor from one command, with a security scanner and a content-hashed lockfile.

toolkit-ai

toolkit-ai treats AI coding-assistant tooling like software dependencies: each assistant keeps its own configuration directory and format for skills, subagents and MCP servers, so toolkit-ai points at GitHub or Bitbucket repositories under the user's control, lets them browse everything in one terminal UI, and installs across all five supported tools, Claude Code, Codex, Amp, GitHub Copilot and Cursor, at once.

A security scanner blocks curl-to-shell patterns, reverse shells, invisible Unicode injection, SSRF and path traversal before install. An interactive terminal UI built with React Ink provides search, filters and multi-select browsing and installing, and a content-hashed lockfile lets a check command show what changed and an update command apply it. The tool ships as a single bundled executable with no runtime dependencies, runnable through npx without cloning.

Built with TypeScript in strict mode, with unit and integration tests typechecked on Node 20 and 22. MIT licensed.

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.