What is MCP?
The Model Context Protocol (MCP) is a standard that allows AI models (like Claude or Gemini) to connect to external tools and data.
In Simple Terms: It gives the AI "arms and legs" to check your work. Instead of just telling you how to deploy, it can run verification commands to see if you did it correctly.
How to Configure
You need an MCP-compatible client. We recommend Antigravity, but you can use any MCP client (like Claude Desktop).
Antigravity Configuration
Antigravity often detects MCP servers automatically or allows you to manage them via the "MCP Servers" panel.
If you need to manually configure it, ensure the server is running and point Antigravity to it.
Claude Desktop / Other Clients
1. Open your client's config file (e.g., claude_desktop_config.json):
2. Add the following configuration (replace the path with your actual path):
{
"mcpServers": {
"cloud-run-tutor": {
"command": "node",
"args": ["/Users/jasondavenport/GitHub/cloud-run-tutor/dist/index.js"]
}
}
}
Troubleshooting
If the AI doesn't see the tools:
- Check the MCP Panel: Ensure "cloud-run-tutor" is listed and running in Antigravity's MCP Servers panel.
- Restart: Try reloading the window or restarting Antigravity.
- Rebuild: Make sure you ran
npm run buildin the project folder first!