Provides submittal, orchestration, and monitoring of automated code reviews via the MAGI gateway.
Configuration
View docs{
"mcpServers": {
"magi_mcp": {
"command": "python",
"args": [
"-m",
"src.server"
],
"env": {
"MAGI_URL": "ws://127.0.0.1:8000/ws",
"DEBUG": "1"
}
}
}
}You run an MCP (Model Context Protocol) server to manage code review submissions and monitor progress with an automated, multi-agent review workflow. This server acts as a bridge between your code submission clients and the MAGI Gateway, coordinating reviews from multiple agents and returning a majority-based decision to help you assess code quality efficiently.
Submit code for review through an MCP client that connects to the MAGI MCP Server. The server forwards your submission to the MAGI Gateway, distributes the review across three agents, and returns the final decision based on the majority of agent votes. You can observe the review progress and results through the client or any integrated dashboards that consume MCP events.
Prerequisites come with Python 3.11 or newer and access to the MAGI Gateway. Optionally, you can deploy in a container using Docker.
Install steps to run the MCP server locally:
1. Install Python dependencies
pip install -r requirements.txtRuntime configuration is centered around connecting to the MAGI Gateway. By default the server connects to ws://127.0.0.1:8000/ws, but you can override this by setting the MAGI_URL environment variable. The MAGI System official gateway is ws://magisystem.ai/ws.
Runtime commands and debugging options shown in examples include running the server directly and enabling debug logging. Docker deployment is also supported for containerized setups.
Orchestrates the code review process by sending submissions to the MAGI Gateway and aggregating agent feedback.
Manages communication with the MAGI Gateway and distributes tasks to review agents (Melchior, Balthasar, Casper).