Rails Hyperdrive MCP
A dev-only Rails engine that plugs an agent into the running app: eight MCP tools plus skills matched to the gems in your Gemfile.
- 82 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The `run_ruby` tool executes arbitrary Ruby inside the running app process.
- The agent reads the database, models and development logs.
Install
- Works in
Manual install
bundle add rails-hyperdrive --group=developmentFollow with `bin/rails hyperdrive:init`. Optionally add a companion gem such as `bundle add rails-hyperdrive-sidekiq --group=development`.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The `run_ruby` tool executes arbitrary Ruby inside the running app process.
- The agent reads the database, models and development logs.
This is third-party code. Review the repository files before installing.
What it does
Rails Hyperdrive mounts an MCP server in your app at `http://localhost:3000/_hyperdrive/mcp`. Eight tools answer from the running process: `run_ruby` executes Ruby, `run_sql` runs read-only queries (up to 100 rows), `tail_logs` reads the log tail, `list_models` and `list_routes` describe models and routes, `locate_source` finds a file and line, `lookup_doc` opens RDoc, and `describe_app` shows versions and gems. The second half: `bin/rails hyperdrive:init` finds skills, guidelines, agents and commands shipped by companion gems and installs only those matching your Gemfile. The gem itself carries no content, it is the mechanism, and knowledge arrives from gems named `rails-hyperdrive-<name>`.
Who it is for. Rails developers who work on an app with Claude Code and want the agent to see the live schema, routes and logs instead of guessing from source.
Good fit when
- The agent picks the wrong association, so you give it `list_models` to see columns and validations.
- You need to find which controller serves a route without paging through files.
- A Sidekiq background job keeps failing and the agent needs the log tail and idempotency rules.
- A team adds a gem with stack-specific rules and wires it up with one command.
Not a fit when
- The app is not running in development: the engine refuses requests outside `Rails.env.development?`.
- You need access to production data, and the tool is built for a local environment only.
- The agent must not execute arbitrary Ruby inside the app process.
Example request
Through Hyperdrive show the Order model with its associations, then the routes that use it and the latest errors from the development log.MCP
- Transport
- http
- Authentication
- not required
Limitations
The engine is development-only: it answers only in `Rails.env.development?`, accepts origins `localhost`, `127.0.0.1` and `[::1]`, and every tool re-checks the mode on invocation. The `run_ruby` tool executes arbitrary code inside the loaded process, so connect it to an app with test data. The gem carries no skills or guidelines itself: without companion gems you get only the MCP tools. The generator writes the `/_hyperdrive` route into `config/routes.rb`, and the path can be changed with the `--mount-at` flag.
How to disable. Remove the gem from the Gemfile, delete the route line from `config/routes.rb`, the `.mcp.json` and `.hyperdrive/` files, and the installed skills and guidelines from `.claude/`.
FAQ
How do I set up Rails Hyperdrive?
Add the gem with `bundle add rails-hyperdrive --group=development`, run `bin/rails hyperdrive:init` and start the server. Claude Code opened in the project directory reads `.mcp.json`.
What does the stack-profile resource return?
The `hyperdrive://stack-profile` resource returns a JSON snapshot of your stack, and `hyperdrive://skills/{name}` returns the body of an installed skill.
Where do I get knowledge for my stack?
Add a companion gem such as `rails-hyperdrive-sidekiq` and run `bin/rails hyperdrive:init` again. It installs only what fits your Gemfile.
Can I change the server address?
Yes, the `--mount-at /path` flag moves both the route and the URL in `.mcp.json`. The path consists of segments separated by `/`.
Does Hyperdrive work in production?
No. The engine answers only in `Rails.env.development?` and accepts origins `localhost`, `127.0.0.1` and `[::1]`.
Discussion
No questions or reviews yet.
Related
MCP serversA Rails gem with an MCP server: the agent reads your schema, models, routes and tests instead of guessing them from a template.
An MCP server and plugin for agents that generate web APIs from OpenAPI, AsyncAPI, GraphQL, gRPC and SQL specs on a Rust core.
The RubyLLM skill and plugin help an agent write chats, tools, MCP and structured output in Ruby and Rails applications
Apollo's MCP server: turns chosen GraphQL operations into tools the agent uses to work with your API
Rebase MCP server: the agent edits the schema, documents and users of a Postgres backend, with writes limited to localhost
The MCP server of the bunqueue job queue: an agent schedules cron jobs, retries failures, sets limits and reads stats, no Redis.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Rails Hyperdrive (this card) | no | not required | 82 | |
| Rails MCP Server | yes | not required | 573 | offers MCP access to a Rails project; Hyperdrive runs inside the live app and matches skills to gems |
| Layered Rails Skills | no | 479 | a skill set only; Hyperdrive adds a live MCP server and knowledge installation from companion gems |
README in short
A development engine for Rails: an MCP server inside the app plus an installer for knowledge from companion gems. Content is not part of the core.