Because the ecosystem of AI-powered developer instruments—from agentic platforms like Antigravity to command-line interfaces like Gemini CLI—continues to increase, a important problem has emerged: how can we guarantee these fashions have entry to essentially the most correct, up-to-date documentation?
Massive Language Fashions (LLMs) are solely nearly as good because the context they’re given. When constructing with Google know-how, builders want their AI assistants to know the most recent Firebase options, the latest Android API modifications, and the present greatest practices for Google Cloud.
As we speak, we’re excited to announce the general public preview of the Developer Information API and its related Mannequin Context Protocol (MCP) server. Collectively, these instruments present a canonical, machine-readable gateway to Google’s official developer documentation.
What’s the Developer Information API?
The Developer Information API is designed to be the programmatic supply of fact for Google’s public documentation. As a substitute of counting on probably outdated coaching information or fragile web-scraping, builders can now search and retrieve Google developer documentation pages as Markdown.
Key options embody:
- Complete protection: Entry documentation from firebase.google.com, developer.android.com, docs.cloud.google.com, and extra.
- Search and retrieve: Discover related documentation pages and snippets, after which retrieve the total Markdown content material.
- Freshness: Throughout our public preview documentation is re-indexed inside 24 hours of an replace, making certain your instruments keep present with the most recent releases.
Powering AI instruments with an MCP server
Alongside the API, we’re releasing an official Mannequin Context Protocol (MCP) server. MCP is an open commonplace that allows AI assistants to securely and simply entry exterior information sources.
By connecting the Developer Information MCP server to your IDE or AI assistant, you give it the power to “learn” Google’s developer documentation. This allows extra dependable options, reminiscent of:
- Implementation steering: “What’s one of the best ways to implement push notifications in Firebase?”
- Troubleshooting: “Are you able to test the docs to learn the way to repair the ApiNotActivatedMapError within the Maps API?”
- Comparative evaluation: “Examine Google Cloud Run and Cloud Features for this particular use case.”
The server is appropriate with a variety of well-liked assistants and instruments, as described within the documentation.
Getting began
You possibly can start utilizing the Developer Information API and MCP server at present in public preview.
- Create an API key: You possibly can generate and prohibit an API key particularly for the Developer Information API inside the Credentials web page of your Google Cloud challenge.
- Allow the MCP server: Set up the Google Cloud CLI, after which allow the MCP server by way of gcloud:
gcloud beta companies mcp allow developerknowledge.googleapis.com --project=PROJECT_ID - Configure your instrument: Replace your instrument’s configuration file (reminiscent of
mcp_config.jsonorsettings.json). Detailed configuration steps for varied AI assistants will be discovered within the documentation.
What’s subsequent?
This preview launch focuses on offering high-quality, unstructured Markdown. As we transfer towards common availability, we plan so as to add help for structured content material reminiscent of particular code pattern objects and API reference entities. We are going to increase the corpus to incorporate extra of Google’s developer documentation and cut back re-indexing latency.
We will’t wait to see the way you combine official Google data into your agentic workflows and developer instruments. Take a look at the full documentation to dive deeper, and tell us what you construct!
