Google has announced the public preview of the Developer Knowledge API. It comes with a Model Context Protocol (MCP) server. This gives AI development tools a simple, machine-readable way to reach Google’s official developer documentation.
The announcement tackles a common issue in AI development. Language models trained on fixed documentation will quickly fall out of sync with fast-changing platforms. The ecosystem of AI-powered developer tools is growing. This includes platforms like Antigravity and command-line tools like Gemini CLI. So, making sure these models have accurate and up-to-date documentation is a big challenge. When AI assistants confidently generate code against deprecated APIs or missing features, the resulting bugs can be subtle and costly to debug.
The Developer Knowledge API acts as a programmatic source of truth for Google’s public documentation.
The API has two main functions.
- SearchDocumentChunks: This finds page URIs and content snippets based on a query.
- GetDocument or BatchGetDocuments: These retrieve the full content of the search results.
Alongside the API, Google is releasing an official MCP server. MCP is an open standard that many in the industry are adopting. It lets AI assistants safely access external data sources in real-time instead of just using their built-in training knowledge. The server offers tools for information retrieval. The search_document tool lets an agent query the documentation using natural language. Meanwhile, get_document retrieves the full content of a specific page found through the search. The practical upshot is that an AI assistant can now look up the authoritative answer to questions like “how do I implement vector search in Firestore?” rather than hallucinating a plausible-sounding but incorrect one.
The MCP server is a remote server, accessible at https://developerknowledge.googleapis.com/mcp. Developers connect to it by enabling the Developer Knowledge API in their Google Cloud project, creating an API key, and updating their tool’s MCP configuration file. Google has published configuration instructions for several popular AI assistants and IDEs.
The current preview release returns documentation as unstructured Markdown. As Google approaches general availability, it will add support for structured content. This includes specific code sample objects and API reference entities. It also plans to expand the documentation and reduce re-indexing latency.
The release fits into a broader pattern of MCP adoption across the industry. This suggests that MCP is becoming the standard way to connect AI agents to live data sources. It’s similar to how REST became the go-to for HTTP APIs a decade ago.
For teams using Google’s developer platforms, the benefit is clear. AI code assistants, which once suggested outdated SDK methods or incorrect configuration options, now have a live reference to consult. This helps bridge the gap between what the model “knows” and what the platform actually supports.
Google’s launch is important not for its uniqueness, but because it rounds out the picture. Now, all three major cloud providers have official, remotely hosted MCP servers. These servers help keep AI coding assistants in sync with their live documentation. AWS’s Knowledge MCP Server became generally available. It offers documentation, blog posts, and Well-Architected guidance without needing authentication. Microsoft’s Learn MCP Server also offers unauthenticated access to the same index that supports Copilot for Azure. It refreshes gradually with each content update. Google’s offering adds API key authentication but still promises sub-24-hour re-indexing after platform updates.
Real-time documentation is quickly becoming a standard expectation for AI tools aimed at developers, not just a feature that sets them apart. What began as separate experiments with MCP has now become a common standard. Each provider is creating a similar “authoritative source of truth” endpoint and linking their AI assistant systems to it. A more exciting competitive edge is the next level up. AWS and Microsoft have gone beyond simple knowledge retrieval. They now offer MCP servers that can act on cloud resources. These servers execute API calls and manage multi-step workflows for agents. Whether Google follows suit with an operational counterpart to its knowledge-focused API will be worth watching as the space continues to mature.
The Developer Knowledge API is available via the Google Cloud console. Detailed setup documentation can be found at developers.google.com/knowledge.
