Back to directory

grimoire

A package manager for AI-agent configuration that installs skills, rules, agents and MCP servers into Claude Code, Copilot, Cursor, Codex, Gemini, Zed, Amp, Kiro, Junie and opencode, storing packages on any OCI registry with no separate service to run.

grimoire

grimoire declares a skill, rule, agent or MCP server once and writes it into whichever coding agents and editors a user has, each in that client's own format, pinned by digest in a lockfile. Storage is any OCI registry, such as GHCR, Docker Hub or a private one, so there is no Grimoire service to sign up for, and where a client has no honest way to receive an artifact, grim says so and skips it rather than writing configuration that looks installed but does nothing.

grim init creates a grimoire.toml, grim add declares, locks and installs a package from a registry reference, grim install re-materializes everything after a fresh clone, and a terminal UI browses the index. A search index is a plain directory of pointers into registries that can be scaffolded and hosted on GitHub or GitLab Pages with one command, giving a searchable catalog and a contribution gate that refuses a pull request claiming a namespace it does not own, again with no index server, database or account required.

Built with Rust, installed through a one-line shell or PowerShell installer, prebuilt binaries, a GitHub Action, or cargo install. Apache-2.0 licensed; stabilizing toward a 1.0 release.

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.