Code Review MCP Server

An MCP (Model Context Protocol) server that executes shell commands for code review. This server provides a tool that can be used by AI assistants to review code or plans by executing the cursor-tools repo command.

Features

• Provides a code review tool that can be used by AI assistants
• Implements the Model Context Protocol for seamless integration with AI assistants
• Uses cursor-tools to perform code reviews
• Enhanced error handling and logging with Winston
• Configurable cursor-tools path
• Input validation and sanitization

Prerequisites

• Bun - JavaScript runtime and package manager
• cursor-tools - Command-line tool for code review

Installation

1. Install dependencies:

bun install

2. Ensure cursor-tools is installed and available in your PATH or configure its location (see Configuration section)

Usage

To run the MCP server:

bun run index.ts

Or use the provided launch script:

./launch-mcp.sh

Configuration

The server can be configured using environment variables:

Example:

CURSOR_TOOLS_PATH=/usr/local/bin/cursor-tools bun run index.ts

Error Handling

The server includes comprehensive error handling:

• Input validation: Empty or invalid inputs are rejected with appropriate error messages
• Command execution errors: Errors during command execution are logged and returned with descriptive messages
• Server startup errors: Issues during server startup are logged with detailed information

All errors are logged to:

• Console output
• error.log (error level messages)
• combined.log (all messages)

Security Considerations

This server executes shell commands based on input received from AI assistants. While basic input sanitization is implemented, be aware of the following:

• The server is designed to be used in a trusted environment
• Input is sanitized to prevent basic command injection, but additional security measures may be needed for production use
• The cursor-tools command should be properly secured and have appropriate permissions

Integration with Roo

{
  "mcpServers": {
    "code-review-server": {
      "autoApprove": [],
      "disabled": false,
      "timeout": 60,
      "command": "bun",
      "args": [
        "run",
        "index.ts"
      ],
      "env": {},
      "transportType": "stdio",
      "alwaysAllow": [
        "review_code"
      ]
    }
  }
}

Testing

To run the test suite:

bun run test.ts

The test suite includes:

• Basic code review functionality
• Complex code review scenarios
• Error handling tests

Logging

Logs are stored in:

• error.log - Contains error-level messages
• combined.log - Contains all log messages (info, warn, error)