Integrations Overview

Pointa integrates with AI coding tools via the Model Context Protocol (MCP). This section covers setup for each supported tool.

Supported AI Tools

Tool Setup Complexity Auto-Start Recommended
Cursor Easy Yes ✅ Best experience
Claude Code Easy Yes ✅ Great for CLI users
Windsurf Easy Yes Good alternative
VS Code Moderate Varies Depends on extension

How Integration Works

All integrations use MCP (Model Context Protocol) to connect:

Your AI Tool  ←→  MCP Protocol  ←→  Pointa Server  ←→  Chrome Extension

The Pointa server exposes tools that your AI can call:

  • read_annotations - Get pending annotations
  • mark_annotations_for_review - Mark work as complete
  • read_issue_reports - Get bug reports
  • And more...

Setup Methods

Auto-Start (Recommended)

Most tools support auto-starting the server:

{
  "mcpServers": {
    "pointa": {
      "command": "npx",
      "args": ["-y", "pointa-server"]
    }
  }
}

Benefits:

  • Server starts automatically when your AI tool opens
  • Always uses the latest version
  • No manual management needed

Manual Start

Some tools require starting the server first:

npx pointa-server start

Then configure the URL endpoint:

{
  "mcpServers": {
    "pointa": {
      "url": "http://127.0.0.1:4242/mcp"
    }
  }
}

Backend Logging (Optional)

For full-stack debugging, run your dev server via pointa dev to capture backend logs:

npx pointa-server dev npm run dev

See Backend Logging for details.

Quick Links