Loading ToolBench

Reading the current corpus…

mcp-nsi-bg MCP server · C · ToolBench

mcp-nsi-bg

Bulgarian National Statistical Institute (НСИ / NSI) — Open Data API. Keyless access to NSI's open datasets via JSON-stat 2.0 format.

2026-07-28+3 tools

69/100

Definition Quality83

Protocol Readiness35

SupportabilityNot measured-

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

Three well-named, read-only tools with clear descriptions and complete input schemas. Tool names follow verb_noun pattern (list_datasets, get_dataset, get_fields). Descriptions are substantive (150 - 200 chars) and explain purpose, prerequisites, and use cases. All parameters have type definitions and descriptions. However, output schemas are not documented, LLMs cannot predict response structure for planning. Error handling is minimal (generic error messages without recovery guidance). No parameter constraints beyond enums for 'lang'. Missing idempotent/readOnly hints in tool metadata.

Tools (3)

get_datasetread onlysource verified83/100

Fetch a single NSI Bulgaria dataset by numeric id as a JSON-stat 2.0 object (dimensions, categories, and the flat `value` array). Get ids from list_datasets. Example ids: 107 (children in kindergartens by municipality), 242 (aggregate replacement ratio).

get_fieldsread onlysource verified83/100

Fetch the field/dimension metadata for an NSI dataset (its column codes plus Bulgarian and English names) as parsed CSV rows. Useful for interpreting the dimension codes in a get_dataset response.

list_datasetsread onlysource verified83/100

List NSI Bulgaria open-data datasets with their numeric id and English (or Bulgarian) name. Each dataset id can be passed to get_dataset for the actual figures. Use the optional `filter` to substring-match dataset names (case-insensitive), e.g. 'population', 'export', 'inflation'.

Issues (5)

HIGHtool ↗

Output schemas not documented. LLMs cannot predict response structure (JSON-stat 2.0 format, CSV rows, etc.) for downstream planning or field extraction.

list_datasetsget_datasetget_fields

HIGHrecovery-guide ↗

Error messages lack recovery guidance. Generic 'NSI: dataset id "X" returned no data' does not tell LLM whether to retry, search for a different ID, or abort.

get_datasetget_fields

MEDIUMtool-annotations ↗

Tool metadata missing readOnlyHint annotations. All three tools are read-only and idempotent, but this is not declared in the tool definition.

list_datasetsget_datasetget_fields

MEDIUMpaginated-result ↗

list_datasets lacks pagination. No limit or offset parameters; response count unbounded. Large result sets could exhaust context window.

list_datasets

Recommendations

Document output schemas for all three tools. For list_datasets, specify: {count: number, lang: 'en'|'bg', datasets: [{id: string, name: string}]}. For get_dataset, reference JSON-stat 2.0 spec and highlight key fields (dimensions, categories, value array). For get_fields, specify CSV row structure.
Add readOnlyHint: true to all tool definitions in McpToolDefinition. This signals to clients that these tools are safe to call without user confirmation.
Enhance error messages with recovery guidance. E.g., 'Dataset id "999" not found. Call list_datasets to discover valid IDs.' or 'Dataset returned empty. This may indicate a deprecated or restricted dataset.'
Add optional limit and offset parameters to list_datasets (default limit=50, max=200). Return total count and next_offset in response to enable pagination.
Document parameter format constraints in descriptions. E.g., 'id: Numeric dataset identifier (1 - 9999, e.g. "107"). Validate with /^\d+$/ before calling NSI API.'

Score history

Overall score trend

First recorded score · v2 rubric

69/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	C	69	2026-07-28+	v2