Seoul Open Data MCP demonstrates solid definition quality with well-structured tool schemas and comprehensive descriptions. All 5 tools have explicit JSON Schema definitions with typed parameters and detailed descriptions (avg 180+ chars). Tool names follow verb_noun convention (recommend_, search_, get_, refine_, list_). However, output schemas are not documented in the visible code, parameter descriptions lack explicit format/constraint guidance in some cases, and error handling recovery paths are not specified. The server targets a specialized domain (Seoul public data) with Korean-language descriptions that are contextually appropriate but reduce LLM cross-language compatibility.

• Add explicit JSON Schema output definitions for all tools. Document the structure of 'recommendations' array (fields: title, provider, type, updateCycle, reason, score, detailUrl) and search result objects.
• Convert parameter descriptions to English or provide bilingual descriptions. At minimum, add English translations in parentheses so non-Korean LLMs can understand intent.
• Define enum constraints for 'division' parameter across all tools: ['본청', '산하기관', '자치구']. Add explicit min/max and defaults for numeric parameters (limit: min=1, default=10).
• Document error scenarios and recovery paths. E.g., 'If no results found, try: (1) broaden keyword, (2) remove orgName filter, (3) call list_seoul_recent_updates to discover available datasets.'
• Add tool annotations (readOnlyHint: true) to all tools since they are read-only queries against Seoul public data API. This signals to agents that these calls are safe to retry.

Inferred effective spec: 2026-07-28+.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry