A server for searching European Union law. Use the `expert_search` tool for broad searches or access a specific document via the `resource://eurlex/document/{celex_number}` resource.
Single tool with clear naming (expert_search) and complete input schema. Description is adequate (145 chars) but lacks actionable guidance on when to use vs. the resource endpoint. Parameters are well-typed with defaults and descriptions. Output schema is inferred from code (list of dicts with celex, title, url) but not formally documented. Error handling returns structured errors but lacks recovery guidance. Resource endpoint (get_document_by_celex) is present but not registered as a tool, limiting discoverability.
Performs an expert search for EU legal documents and returns a clean list of results. The query must be written using EUR-Lex expert query syntax. Example query: 'DN = 32024R*' to find all regulations from 2024.
Output schema not formally documented. Code infers list[dict] with celex/title/url fields, but LLMs cannot see this structure in the tool definition. Agents must guess what fields to extract.
Description lacks actionable guidance on when to use expert_search vs. the resource://eurlex/document/{celex_number} endpoint. LLMs cannot distinguish between the two approaches without explicit direction.
Error handling returns {'error': str(e)} but does not guide recovery. When a query fails, the LLM receives no actionable next step (e.g., 'Try a simpler query' or 'Check EUR-Lex syntax').
Resource endpoint get_document_by_celex is implemented but not exposed as a discoverable tool. Agents must infer the resource URI pattern from server instructions rather than seeing a formal tool definition.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2026-07-28+ | v2 |
No pagination metadata returned. Results include page/page_size params but response does not include total_count or has_more, preventing agents from knowing if more results exist.