In September, 2025, we introduced the Data Commons Model Context Protocol (MCP) server to provide a standard way for AI agents to interact with Data Commons data natively. This was followed by the Gemini CLI extension, which streamlined the setup process for Gemini CLI users.
While these tools successfully opened the door for natural language data exploration, they relied on local Python environments, which created some challenges. Installing and running open-source tools was not compatible with high-security environments; and hosting a local MCP server lacked the scalability needed for developers who want to publish their own query agents.
Today, we’re taking the next step to make it easier for everyone to use datacommons.org with an LLM by launching the hosted Data Commons MCP service in Google Cloud Platform. Now you can just connect to the free service and let Google manage the rest – no need to worry about Python environments, resource management, version releases, security compliance etc.
The MCP server provides a standardized way for AI agents to consume Data Commons data natively. Analysts can create insights by asking high-level questions in natural language with the MCP returning data from trusted sources, while developers can easily create AI agents customized for their needs. For example, you can get statistical answers to questions like "What is the correlation between unemployment levels and obesity rates in U.S. states?" or "Rank-order the GDP of every eastern European country".
If you're already using the Data Commons Gemini CLI extension, you don't need to do anything. The next time you run Gemini CLI, the extension will automatically update itself to connect to the server over the web instead of starting up a local server instance. If you haven't used the extension before, see Use the Gemini CLI extension to get started.
If you're using Gemini CLI without the extension, or any other agent, you'll need to get a free Data Commons API key and update your configuration to point to the hosted server. For complete details, see Use Gemini CLI.
"mcpServers": {
"datacommons-mcp": {
"httpUrl": "https://api.datacommons.org/mcp",
"headers": {
"X-API-Key": "YOUR DC API KEY"
}
}
}And as always, if you’re interested in sharing your cool usage of Data Commons or have feedback, please email us at support@datacommons.org.
Note: The hosted MCP server can only be used to query datacommons.org. If you're running your own Custom Data Commons instance, you'll still need to run your own MCP server. See Run MCP tools for more information.
