US Census Bureau MCP Server

US Census Bureau

Access US Census Bureau demographic, economic, housing, and geographic data. Covers 1,700+ datasets including the American Community Survey (ACS), Current Population Survey (CPS), Decennial Census, Population Estimates, County Business Patterns, and economic time series. MCPBundles supplies the Census Data API key server-side, so users can enable the server without entering a Census credential.

Workflow

1. Search groups by topic keyword (e.g. "income", "housing") to find table IDs
2. Look up variables in that group to get exact codes (codes ending in E = estimates)
3. Check geographies with common_only=true to see valid levels and for/in clause syntax
4. Query data with the variable codes and geography clauses

For common surveys, skip dataset discovery — use year+survey shortcuts (e.g. year="2022", survey="acs5").

Key Concepts

• Surveys: acs5 (most complete coverage, all geographies), acs1 (65K+ pop areas only), acs5/profile (pre-computed percentages), dec/dhc (decennial), pep/population (estimates), cbp (business patterns)
• Variables: Alphanumeric codes like B01003_001E. Search by keyword. Always include NAME for readable labels.
• Geography: Hierarchical FIPS codes. for_clause targets the level, in_clause constrains the parent. Use the geography tool to get exact syntax — it generates ready-to-use for/in clause templates.
• FIPS lookup: Don't know a FIPS code? Query data with get_variables=NAME and for_clause targeting the level (e.g. county:*) to get names alongside codes.
• Sorting: Data queries support sort_by (any variable code) + sort_order for ranked results.
• Variable types: Defaults to estimates only. Use variable_type="moe" for margins of error, "all" for everything.

Gotchas

• ACS 1-year excludes areas under 65K population — use acs5 for small geographies
• Sub-state queries (county, tract, place) require in_clause with state FIPS
• Profile tables use DP-prefixed variables, subject tables use S-prefixed — search variables within the right survey