Statistics Canada MCP Server

Each tool targets a distinct facet of the StatCan workflow: release tracking (get_changes), observations (get_sdmx_data), dimension codes (get_sdmx_structure), series metadata (get_series_info), and table discovery (search_cubes_by_title). The closest potential confusion is between get_sdmx_structure and get_series_info, since both describe a table/series, but the descriptions clearly separate key dimensions from metadata fields.

All names are snake_case with a leading verb (get_/search_), which gives a predictable read. Minor deviation: two tools carry a domain prefix (get_sdmx_*) while the others (get_changes, get_series_info, search_cubes_by_title) do not, but the overall pattern is still clear.

Five tools is on the lean side but each maps to a real step in the workflow (find table -> inspect structure -> fetch data/metadata -> check updates). Nothing feels redundant, though a small addition like series/vector search would make the set feel less thin.

The surface covers discovery, schema inspection, data retrieval with server-side SQL, metadata, and release tracking—solid coverage for a statistics data API. A minor gap is the inability to search series/vectors directly by name; agents must derive vectorIds via key_template plus structure lookups, which is workable but indirect.