The Firebase MCP server is a local Model Context Protocol server that runs on your development machine. It lets AI-powered development tools work with your Firebase projects and your app's codebase from inside your editor.
An editor set up to use the server can help you create and manage Firebase projects, manage Firebase Authentication users, work with data in Cloud Firestore and Firebase SQL Connect, retrieve SQL Connect schemas, understand security rules for Cloud Firestore and Cloud Storage, and send messages with Firebase Cloud Messaging.
The server works with any tool that can act as an MCP client over standard I/O, including Antigravity, Claude Code and Claude Desktop, Cline, Cursor, VS Code Copilot and Windsurf. Setup requires a working installation of Node.js and npm. The page gives configuration steps for several of these editors, using the npx command to run firebase-tools.
When the server makes tool calls, it uses the same user credentials that authorise the Firebase CLI, either a logged-in user or Application Default Credentials. Some tools, such as certain Cloud Firestore tools, send requests to a remote MCP server or backend service to reach cloud-hosted resources.
The page also suggests pairing the server with Firebase agent skills, which teach AI agents Firebase best practices for tasks such as setting up Authentication, Cloud Firestore or Firebase AI Logic, and deploying to Firebase Hosting or App Hosting.
The page describes the server and its setup but does not list a separate price for it. Firebase itself has a no-cost Spark plan with usage limits and a pay-as-you-go Blaze plan that includes the Spark usage.
