Python MCP servers exposing Open Data to compatible LLM clients.
OpenDataMCP exposes 2 tools for Swiss railway data (rail-traffic-info, railway-lines). Both tools have complete, well-structured input schemas with reasonable parameter descriptions and constraints (min/max, defaults, enums). However, the tool descriptions are generic ('Fetch rail traffic information', 'Fetch railway line information') and lack WHEN-to-use guidance or differentiation. Descriptions fall in the 30-40 character range, below the 50-200 char baseline for LLM-optimized prompts. Parameter descriptions are present and actionable (e.g. 'Fields to select', 'Filter conditions'), but output schemas are not documented, the response structure is not visible in the source, making it impossible to verify what fields agents should expect. No error handling guidance is visible. Tool names are clear (verb_noun pattern) but somewhat generic. No tool annotations (readOnlyHint, etc.) are present despite both being read-only operations.
Fetch rail traffic information
Fetch railway line information
Tool descriptions are too generic and lack WHEN-to-use context. 'Fetch rail traffic information' does not distinguish this tool from railway-lines or guide LLM selection. Baseline for LLM-optimized descriptions is 50 - 200 chars; both tools are ~35 chars.
Output schemas are not documented. The source code does not show what fields rail-traffic-info or railway-lines return. LLMs cannot plan downstream tool calls or extract required fields (e.g. channel_id for follow-up operations) without knowing the response structure. This violates pattern:tool.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Both tools are read-only; annotating them would signal to agents they are safe to invoke without confirmation. Missing current MCP spec capability.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 53 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 42 | - | v1 |
No error handling guidance visible. If a query filters for non-existent traffic or railway lines, what is returned? How should the LLM interpret an empty result vs. an API error? Pattern:recovery-guide requires error responses to tell the LLM what to do next.