Back to directory

Rails AI Agents

Rails AI Agents is a production-ready Claude Code configuration for Ruby on Rails development, bundling 19 specialized agents, 26 slash commands, 18 skills, 14 rules and 6 lifecycle hooks that teach the assistant Rails conventions and TDD workflows.

Visit website (opens in a new tab)thibautbaissac.github.io
Rails AI Agents

Rails AI Agents is copied directly into a Rails project's .claude/ directory to give Claude Code Rails-specific agents, commands, skills, rules and hooks, including a Spec Driven Development kit that runs a full specification-to-implementation pipeline as well as a lightweight mode for bug fixes. An alternative conventions pack favors a 37signals-style approach with rich domain models and Minitest instead of the default layered architecture with RSpec and Pundit.

It also supports OpenAI Codex and GitHub Copilot alongside Claude Code, keeping .claude/ as the canonical source of truth: an AGENTS.md file gives shared high-level instructions, a mirrored skills directory serves Codex and Copilot, and generated path-scoped bridge files point back to the canonical Claude rules, with a local sync script working around unreliable Codex skill discovery through symlinks.

Additional pieces include a status line, a Sentry MCP integration, and plan, review and pull-request artifact commands.

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.