Server has 5 tools with complete input schemas and descriptions. Naming follows verb_noun pattern (normalize_, get_, resolve_, enrich_). Descriptions are domain-specific and contextual (avg ~150 chars), exceeding the 10-char minimum. However, parameter descriptions lack actionable constraints (ranges, formats, enums). Output schemas are partially documented via Pydantic models but not explicitly declared in tool registration. Error handling is minimal, no recovery guidance or categorization. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite all tools being read-only. Missing per-parameter validation rules and dependency documentation.

• Add readOnlyHint=true to all 5 tools via @mcp.tool(annotations={...}) to signal safety to agents.
• Expand parameter descriptions with format/range guidance: 'max_candidates (1-20, default 5): limits returned address candidates to reduce token usage'.
• Document output schemas in tool descriptions: 'Returns NormalizeResult with fields: normalized (dict|null), candidates (list of dicts with keys: road_addr, jibun_addr, postcode5, etc.)'.
• Add recovery guidance to error responses: 'If road_addr and jibun_addr both missing, call normalize_address() first to extract a standard address.'
• Clarify kakao_place vs kakao_places precedence in resolve_from_kakao_place: 'If both provided, kakao_place takes precedence. Provide only one.'

Inferred effective spec: <=2025-11-25.

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