Loading ToolBench
INE Spain (Instituto Nacional de Estadística) Tempus3 JSON API MCP. Keyless access to Spain's official statistical time-series API — CPI (IPC), labor force survey (EPA), housing, industry, etc.
INE Spain MCP has well-structured tool definitions with clear naming (verb_noun pattern), comprehensive descriptions (avg 180 chars), and complete input schemas with type constraints. All 6 tools follow consistent patterns. However, output schemas are not documented, LLMs cannot predict response structure for downstream chaining. Parameter descriptions are good but lack format/range guidance (e.g., 'YYYYMMDD:' format not formally specified). No error handling guidance or recovery paths documented. No tool annotations (readOnlyHint, etc.) despite all tools being read-only. Missing pagination guidance for potentially large result sets.
List all available INE statistical operations (Spain). Each has a numeric Id, a Codigo (e.g. "IPC"=CPI, "EPA"=labor force survey), and a Nombre. Use the Codigo or Id with tables_for_operation.
Fetch data for ONE INE series by its code. Use nult=N for the last N periods, or date="YYYYMMDD:" for from that date on. Returns the series with its Data[] array of {Fecha (epoch ms), Anyo, Valor} points.
List the data series contained in an INE table. Each series has a COD (series code, e.g. "IPC251852") and a Nombre describing the breakdown. Pass a COD to series_data.
Fetch data points for ALL series in an INE table at once. Use nult=N for the last N periods, or date="YYYYMMDD:" for everything from that date on (trailing colon = open-ended range). Each series object carries a Data[] array of {Fecha (epoch ms), Anyo, Valor}.
List the statistical tables belonging to one INE operation. operationId may be the numeric Id (e.g. 25) or the operation code (e.g. "IPC", "EPA"). Each table has an Id used by series_in_table / table_data.
Output schemas not documented. LLMs cannot predict response structure (e.g., Data[] array format, field types) for downstream tool chaining or result extraction.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All 6 tools are read-only and idempotent, annotate them to help agents optimize retry logic and avoid unnecessary re-execution.
Date/range parameter format not formally specified. 'date="YYYYMMDD:"' is documented in description text but not as a JSON Schema pattern constraint. LLMs may pass invalid formats.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 68 | 2026-07-28+ | v2 |
List the values of an INE classification variable (e.g. provinces, age groups, ECOICOP groups). With no variableId, lists all variables; with a variableId, lists that variable's allowed values (each has an Id, Codigo, Nombre). Useful for understanding the breakdowns inside tables.
No error handling guidance. If an invalid operationId or tableId is passed, the tool throws a generic HTTP error. No recovery hints (e.g., 'Try list_operations() first to find valid IDs').
Pagination not addressed. table_data and series_data may return large result sets (e.g., 50+ years of monthly data). No limit, offset, or cursor parameters documented.