LoopBreaker MCP

Get AI second opinions when stuck debugging. Works with Claude with OAuth authentication.

Quick Setup (30 seconds)

# Get help and see all available commands
npx loopbreaker-mcp --help

# Step 1: Login (optional but recommended)
npx loopbreaker-mcp --login

# Step 2: Setup MCP integration
npx loopbreaker-mcp --setup

Restart Claude and you're ready!

📋 Available Commands

• --help or -h: Show help and usage information
• --login: Open browser for Google OAuth authentication
• --setup: Configure Claude Code MCP integration
• (no args): Start MCP server (stdio mode) - usually automatic

Usage

In Claude, just ask:

• "Use loopbreaker to help debug this error"
• "Get a second opinion on why this isn't working"
• "Ask loopbreaker why this function fails"

What it does

LoopBreaker analyzes your:

• Error messages & stack traces
• Code context
• Recent chat history

Then provides fresh perspectives from multiple AI models to help you break out of debugging loops.

Authentication

LoopBreaker now supports OAuth authentication for enhanced user tracking and personalized debugging assistance:

# Login with Google OAuth
npx loopbreaker-mcp --login

Your authentication token is securely stored in ~/.claude/auth.json.

Manual Setup

If you prefer manual configuration, add to Claude's config:

{
  "mcpServers": {
    "loopbreaker": {
      "command": "npx",
      "args": ["-y", "loopbreaker-mcp"],
      "env": {
        "LOOPBREAKER_API_URL": "https://koi.otonomy.ai",
        "LOOPBREAKER_API_TOKEN": "loopbreaker_prod_2025_secure_koi"
      }
    }
  }
}

Support

• Package: npm package page
• API Status: koi.otonomy.ai/healthz

Made with ❤️ to save your debugging time