Tidewave Rails

io.github.tidewave-aiPublished by GitHub account tidewave-aiv0.8.0Updated Oct 24, 2026

Runtime-level MCP server for Rails development: run code, read logs, query the DB, read docs.

The publisher's own server: Job Bot by Agent Rails

VerifiedSTDIODesktop onlyDatabasesDeveloper Tools

Overview

AI-generated overview

Runtime-level MCP server for Ruby on Rails development that lets an assistant run code in a live app, read logs, query the dev database, and look up docs and源码.

What you can do
Run Ruby code inside the running Rails app and get results plus stdout, like a Rails console for the agent. Read the app's live logs. Run SQL queries against the development database and get rows back. Look up documentation for classes, methods and constants from the exact gem versions in Gemfile.lock. Find the file and line where a class, module or method is defined, including in gem dependencies and metaprogrammed methods.
When to use it
Useful when an agent is working on a Rails app and should ground its changes in actual runtime behavior: reproducing bugs against live state, verifying records after a create action, or checking how a gem method works in the pinned version instead of relying on stale training data.
Requirements
A local Ruby on Rails app with the tidewave gem added to the development group, and the app running in development. The MCP client connects over streamable HTTP to the app's /tidewave/mcp path and port, for example localhost:3000. Desktop MCP client only.
Before you install
The assistant can execute arbitrary Ruby code in the app and run any SQL against the development database, including inserts and other writes. Requests are limited to localhost by default; allow_remote_access should only be enabled knowingly. Tidewave is not meant for production and raises an error where code reloading is disabled. It also relaxes Content-Security-Policy (unsafe-eval, frame-ancestors) in loaded environments.

Installation

In SourceWeft

  1. Open Tidewave Rails in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Tidewave Rails

Tidewave Rails is an MCP server that provides runtime-level tools for developing Ruby on Rails apps using coding agents.

Your agent will be able to use this MCP server to talk to your running Rails app in development to:

  • execute code in the context of the running app (like a Rails console for agents)
  • read the app's live logs
  • query your development database
  • get source locations of classes and methods
  • read documentation pinned to the exact gem versions your app depends on

This MCP server is an open-source component of Tidewave, the agentic development environment for Rails and Phoenix.

You can use this project as a standalone MCP server or integrated with the Tidewave product by following the instructions below.

Installation

1. Add the Tidewave gem to your app

You can add Tidewave Rails to your app by running:

shell
bundle add tidewave --group development

or by manually adding the tidewave gem to the development group in your Gemfile:

ruby
gem "tidewave", group: :development

2. Add the Tidewave MCP to your agent/editor

Add the Tidewave MCP server to your editor or MCP client configuration as the type "http" (streamable), pointing to the /tidewave/mcp path and port your web application is running at. For example, http://localhost:3000/tidewave/mcp.

We also have specific instructions for:

Usage

As with any other MCP server, your agent will call the tools exposed by the Tidewave MCP whenever it sees fit. But you can also prompt it to call them explicitly.

Available MCP tools

project_eval

Evaluates Ruby code in the context of your running application, with access to its runtime, loaded dependencies, and in-memory data, returning the result plus anything printed to standard output. It's like a Rails console for the agent.

[project_eval demo]

Your agent can use it when it would rather run code than assume behavior, grounding its next step in what the running app actually does. For example, calling a method to see what comes back or reproducing a failing code path against live app state to debug it.

execute_sql_query

Runs a SQL query against your app's development database and returns the rows to the agent.

[execute_sql_query demo]

Your agent can use it to run any SQL against your development database. For example, ask it to insert some test records to see how a page looks with realistic data. Or, after a create action, the agent can verify whether the record was saved with the expected values.

get_docs

Looks up the documentation for a class, method, or constant, reading from the exact gem versions locked in your app's Gemfile.lock.

[get_docs demo]

Your agent can use it when it's unsure how a class or method works, so the code it generates is grounded in the docs for the exact versions of the gems your app uses, rather than training data that may be stale or a generic docs lookup that can't guarantee it matches the version your app depends on.

get_source_location

Returns the file and line where a class, module, or method is defined, across both your app and its dependencies.

[get_source_location demo]

Your agent can use it to jump straight to where a class or method is defined, by file and line, instead of grepping for it, including when the definition lives in a gem dependency.

Also, because it resolves the location from your running app instead of parsing source text, it handles metaprogramming, where a method is generated at runtime and does not appear as a literal def for grep to find.

[!NOTE]

Why no tools for routes, associations, etc?

Tidewave does not include tools for listing your routes, associations, etc. because agents are better off reading their respective source files, which gives agents more context and enables them to perform any necessary edit without additional tool calls.

Instead, Tidewave aims to fill in missing gaps, such as evaluating code inside your Rails app (without starting new instances) and finding source location, which can be tricky, even with grepping, due to metaprogramming and the different places Bundler can install your dependencies.

Troubleshooting

The Tidewave toolbar is missing

This may happen if you are compressing your responses (gzip, brotli, etc) after the Tidewave middleware runs. Use bin/rails middleware and make sure Tidewave comes after Rack::Deflater or similar. Also look into your browser and terminal logs for any errors.

Using multiple hosts/subdomains

If you are using multiple hosts/subdomains during development, you must use *.localhost, as such domains are considered secure by browsers. Additionally, add the following to config/initializers/development.rb:

ruby
config.session_store :cookie_store,  key: "__your_app_session",  same_site: :none,  secure: true,  assume_ssl: true

And make sure you are using rack-session version 2.1.0 or later.

The above will allow your application to run embedded within Tidewave across multiple subdomains, as long as it is using a secure context (such as admin.localhost, www.foobar.localhost, etc).

Content security policy

If you have enabled Content-Security-Policy, Tidewave will automatically enable "unsafe-eval" under script-src in order for contextual browser testing to work correctly. It also disables the frame-ancestors directive. This is done only in the environments that Tidewave is loadead (development by default).

Production Environment

Tidewave is a powerful tool that can help you develop your web application faster and more efficiently. However, it is important to note that Tidewave is not meant to be used in a production environment.

Tidewave will raise an error if it is used in any environment where code reloading is disabled (which typically includes production).

Configuration

You may configure tidewave using the following syntax:

ruby
  config.tidewave.team = { id: "my-company" }

The following config is available:

  • allow_remote_access - Tidewave only allows requests from localhost by default, even if your server listens on other interfaces, for security purposes. Read our security guidelines for more information and when to allow remote access (if you know what you are doing)

  • logger_middleware - The logger middleware Tidewave should wrap to silence its own logs

  • preferred_orm - which ORM to use, either :active_record (default) or :sequel

  • team - set your Tidewave Team configuration, such as config.tidewave.team = { id: "my-company" }

  • toolbar - controls whether the Tidewave toolbar is injected into HTML pages. Defaults to true

Acknowledgements

A thank you to Yorick Jacquin for the initial version of this project.

Development

Run the Minitest suite with:

shell
bundle exec ruby -Itest test/all_test.rb

License

Copyright (c) 2025 Dashbit

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Source: README.md at commit 420b171

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.8.0LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

Other Rails integrations