Quick Start

MCP Quick Start

This guide gets you from zero to running architecture analysis via your AI assistant in Cursor.

1. Install the MCP Server

Download the arxo-mcp binary for your platform from GitHub Releases. Place it on your PATH or note the full path for the next step.

2. Configure Cursor

Add the Arxo MCP server to .cursor/mcp.json in your project or user config:

{
  "mcpServers": {
    "arxo": {
      "command": "/path/to/arxo-mcp",
      "args": [],
      "env": {}
    }
  }
}

Replace /path/to/arxo-mcp with the actual path to the binary (e.g. /usr/local/bin/arxo-mcp or the full path to the extracted binary).

3. Restart Cursor

Reload the window or restart Cursor so it picks up the new MCP configuration. The Arxo server will start automatically when the AI needs it.

4. Try It

In the AI chat, ask:

Check for circular dependencies in this project.

The AI will call check_cycles and report any cycles found.

Or try:

Run a quick architecture analysis on this codebase.

The AI will call analyze_architecture with the quick preset and summarize metrics like SCC (cycles), centrality, and more.

5. Verify It Works

If the AI doesn’t seem to use Arxo tools:

1. Check that the binary path in your config is correct
2. Open MCP Server Logs in Cursor and look for the Arxo server
3. See Troubleshooting for common issues

Next Steps

• Tools — All available MCP tools and when to use them
• Workflows — Example AI assistant workflows (policy enforcement, impact analysis, LLM audit)
• Configuration — Cache TTL, log level, and other options