USAFacts Government Data Connector

By USAFacts

Overview

Connect Claude to government data organized and maintained by USAFacts. Ask questions about the state of the nation and get answers centered in real data from dozens of federal sources like the Census Bureau, BLS, and Treasury, with added context to help Claude interpret and explain the numbers. Start any prompt with "use usafacts" to use the connector. Then ask about topics like unemployment, housing, inflation, federal spending, and the national debt. Not sure where to start? Try these prompts: - What's the national debt, and how fast is it growing? - How does unemployment in my state compare to the national rate? - What's actually driving inflation right now... rent, groceries, or something else? - Where does federal spending go, and which agencies spend the most? - How is the population in my state changing? We're starting with economic data and adding more government datasets over time. Learn more about this connector and see what data is currently available: https://usafacts.org/claude-connector/

Tools

call_source_tool

Claude

Call a tool on a data source. Requires a source_id and tool_name previously returned by get_source_tools(), with an arguments object matching the schema it gave for that tool. On success, the response carries _citation -- use it for attribution rather than reconstructing one. Read-only: every source in this gateway (U.S. government statistics APIs, legislative and spending data, .gov web search -- see list_sources() for the full, current list) only retrieves data, even where the upstream requires POST for a request body.

get_source

Claude

Get a source's playbook, methodology, and tool index. The playbook names the tool for each question shape and defines the source's concepts. The tool index carries names, summaries and groups only, not schemas; get_source_tools() returns the schema for named tools from that index.

get_source_tools

Claude

Get the full description and argument schema of named source tools. call_source_tool() requires a schema previously returned here for the source_id/tool_name pair it's given. Name every tool you may call in one request; call this once per source, after get_source() and before any call_source_tool.