MCP Setup

Connect Pointa to your AI coding tool using the Model Context Protocol (MCP). This allows your AI agent to read your annotations and implement changes automatically.

What is MCP?

MCP (Model Context Protocol) is a standard that allows AI coding tools to communicate with external services. Pointa uses MCP to share your annotations, bug reports, and design changes with your AI agent.

When configured, your AI tool can:

  • Read all pending annotations
  • See bug report timelines
  • Access design mode changes
  • Mark work as complete

Quick Setup by Tool

Choose your AI coding tool:

Cursor

Cursor has excellent MCP support and is the recommended way to use Pointa.

Steps

  1. Open Cursor
  2. Go to Settings → Cursor Settings
  3. Navigate to the Tools & Integrations tab
  4. Click + Add new global MCP server
  5. Paste this configuration:
{
  "mcpServers": {
    "pointa": {
      "command": "npx",
      "args": ["-y", "pointa-server"]
    }
  }
}
  1. Click Save
  2. Restart Cursor

Verify Connection

After restarting:

  1. Open the Cursor Agent (Cmd/Ctrl + I)
  2. Look for "pointa" in the available tools list
  3. You should see tools like read_annotations, mark_annotations_for_review, etc.

Claude Code

Claude Code supports MCP via configuration file or CLI.

Option 1: CLI Method

Run this command in your terminal:

claude mcp add --transport http pointa http://127.0.0.1:4242/mcp

Important: With this method, you need to start the server manually first:

npx pointa-server start

Option 2: Config File Method

Add to your Claude configuration file (usually ~/.config/claude/config.json):

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

Verify Connection

Start a new Claude Code session and check that Pointa tools are available.

Windsurf

Steps

  1. Open Windsurf
  2. Go to Settings → Advanced Settings
  3. Scroll to the Cascade section
  4. Add this configuration:
{
  "mcpServers": {
    "pointa": {
      "command": "npx",
      "args": ["-y", "pointa-server"]
    }
  }
}
  1. Save and restart Windsurf

VS Code

VS Code requires an AI extension that supports MCP, such as GitHub Copilot Chat or Continue.

General Steps

  1. Install an MCP-compatible AI extension
  2. Start the Pointa server manually: 
    npx pointa-server start
  3. Configure your AI extension with the MCP endpoint: 
    http://127.0.0.1:4242/mcp

Note: Configuration varies by extension. Check your specific extension's documentation for MCP setup instructions.

Other Editors

For other editors that support MCP:

If your editor supports command-based MCP:

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

If your editor only supports URL-based MCP:

  1. Start the server manually:

    npx pointa-server start

  2. Configure your editor with:

    http://127.0.0.1:4242/mcp

Troubleshooting MCP Connection

Server not detected

Run this command to check the server status:

npx pointa-server status

If the server isn't running, start it:

npx pointa-server start

Tools not appearing in AI

  1. Make sure you saved the configuration
  2. Restart your AI tool completely
  3. Check that the configuration JSON is valid (no syntax errors)

Port already in use

If port 4242 is in use, you can change it:

POINTA_PORT=4243 npx pointa-server start

Note: You'll also need to update the Chrome extension settings if you change the port.

Next Step

Continue to First Annotation to create your first annotation and see AI implement it.