chester-2026/gov-stat-mcp-server

GovData MCP is a server that lets AI agents search, explore, and fetch open datasets from official government data portals. It follows the Model Context Protocol, a standard that lets tools like Claude Desktop or Claude Code plug in outside data sources. Instead of building a separate connection to each country's data site, an agent talks to this one server and gets a single, consistent set of commands. It currently covers four national open data portals: the United States (data.gov), the United Kingdom (data.gov.uk), Canada (open.canada.ca), and Australia (data.gov.au). Each of these portals has its own quirks, such as different login requirements, bilingual titles, or differences in how files are listed, and the server hides those differences behind one common set of tools. The typical flow for an agent is to list the available portals, search for a dataset by keyword, look at a dataset's full details, then either preview a few rows of a file, download it directly, or get a link to download it later. Descriptions are shortened so they do not use up too much of the agent's context, and errors are written as plain instructions rather than raw technical error messages, so an agent can recover and try a different approach on its own. Setup requires Python 3.11 or newer. Using the US portal needs a free API key from api.data.gov, while the other three portals need no account at all. The server can run locally using the stdio method that tools like Claude Desktop expect, or as an HTTP server, which makes it suitable for deployment on a cloud service such as Google Cloud Run. A Docker setup is included for running it that way. This project is aimed at developers building AI agents that need access to public government statistics and datasets, giving them one dependable interface across multiple countries instead of writing custom code for each portal.