About

A Model Context Protocol (MCP) server that enables AI coding assistants to control launch, inspect and mutate Bevy applications via the Bevy Remote Protocol (BRP). This tool bridges the gap between coding agents and Bevy by providing comprehensive BRP integration as an MCP server.

Bevy Compatibility

The bevy_brp_mcp crate follows Bevy's version numbering and releases new versions for each Bevy release.

Features

Core BRP Operations

• Entity Management: Create, destroy, query, and modify entities
• Component Operations: Get, insert, remove, and mutate components on entities
• Resource Management: Access and modify global resources
• Query System: Advanced entity querying with filters
• Hierarchy Operations: Parent-child entity relationships

Application Discovery & Management

• App Discovery: Find and list Bevy applications in your workspace
• Build Status: Check which apps are built and ready to run
• Launch Management: Start apps with proper asset loading and logging
• Example Support: Discover and run Bevy examples from your projects

Real-time Monitoring

• Component Watching: Monitor component changes on specific entities
• Log Management: Centralized logging for all launched applications
• Process Status: Check if apps are running with BRP enabled

Enhanced BRP Integration

requires bevy_brp_extras

• Format Discovery: Get correct JSON formats for BRP operations (via bevy_brp_extras)
• Screenshot Capture: Take screenshots of running Bevy applications
• Graceful Shutdown: Clean application termination

Getting started

first, install via cargo: cargo install bevy_brp_mcp

configure your mcp server - for claude code this would be in the ~/.claude.json file.

"mcpServers": {
  "brp": {
    "type": "stdio",
    "command": "bevy_brp_mcp",
    "args": [],
    "env": {}
  }
},

that's it!

Usage

With AI Coding Assistants

bevy_brp_mcp is designed to be used with AI coding assistants that support MCP (like Claude). The MCP server provides tools that allow the AI to:

1. Discover and launch your Bevy applications - with logs stored in your temp dir so they can be accessed by the coding assistant.
2. Inspect and modify entity components in real-time
3. Monitor application state and debug issues
4. Take screenshots and manage application lifecycle (requries bevy_brp_extras)

Setting Up Your Bevy App

For full functionality, your Bevy app should include BRP support:

use bevy::prelude::*;

fn main() {
    App::new()
        .add_plugins(DefaultPlugins)
        .add_plugins(bevy::remote::RemotePlugin::default()) // Enable BRP
        .run();
}

For enhanced features like screenshots and format discovery, also add bevy_brp_extras:

use bevy::prelude::*;
use bevy_brp_extras::BrpExtrasPlugin;

fn main() {
    App::new()
        .add_plugins(DefaultPlugins)
        .add_plugins(BrpExtrasPlugin) // Enhanced BRP features
        .run();
}

In either case you'll need to make sure to enable bevy's "bevy_remote" feature.

Integration with bevy_brp_extras

This crate is designed to work seamlessly with bevy_brp_extras. When both are used together:

1. Add BrpExtrasPlugin to your Bevy app for enhanced BRP features
2. Use bevy_brp_mcp with your AI coding assistant
3. Additional methods like screenshot, shutdown, and format discovery will be automatically available
4. Get proper JSON formats for complex BRP operations. The brp_extras/discover_format feature is especially useful. The value returned from bevy/registry/schema does not tell you exactly what is expected by the brp spawn/insert/mutate calls. As a result your coding agent will engage in trial and error to figure it out but it's not 100% reliable.

If you have bevy_brp_extras installed, it can get the type information directly from the running app andand provide it if queried via brp_extras/discover_format - or it will provide it in the error message if your coding agent tries a call and fails.

Example Workflow

1. Discovery: Use list_bevy_apps to find available applications
2. Launch: Use launch_bevy_app to start your game with proper logging
3. Inspect: Use bevy_query to find entities of interest
4. Monitor: Use brp_get_watch to observe entity changes in real-time
5. Modify: Use bevy_mutate_component to adjust entity properties
6. Debug: Use read_log to examine application output
7. Capture: Use brp_extras_screenshot to document current state

Logging

All launched applications create detailed log files in /tmp/ with names like:

• bevy_brp_mcp_myapp_1234567890.log (application logs)
• bevy_brp_mcp_watch_123_get_456_1234567890.log (monitoring logs)

Use the log management tools to view and clean up these files.

License

Dual-licensed under either:

• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)

at your option.