Two tools with clear verb-noun naming (list_businesses, get_business) and reasonable descriptions (100-150 chars). Both have documented input schemas with types and descriptions. However, output schemas are not explicitly documented in the code, the tool docstrings describe what is returned in prose but do not provide a formal schema structure. Parameter descriptions are present but lack detail on constraints (e.g., 'sort' accepts three specific values but they are not enumerated; 'limit' range is stated in prose but not as min/max schema constraints). No error handling guidance, no recovery hints, and no tool annotations (readOnlyHint, idempotentHint). The server is read-only and stateless, which is good, but the definitions lack the rigor expected of production tools.

• Add formal JSON Schema output definitions for both tools. Document field names, types, and whether fields are always present or optional. Example: list_businesses returns [{slug: string, name: string, url: string, averageRating: float, reviewCount: int, address: string|null, priceLevel: string|null}, ...]
• Convert 'sort' parameter to an enum: {type: 'string', enum: ['top', 'most-reviewed', 'newest'], default: 'top'}. Add minItems and maxItems to 'limit': {type: 'integer', minimum: 1, maximum: 100, default: 25}.
• Add error handling guidance to tool descriptions. Example: 'If city or category slug is invalid, returns empty list. Call list_cities() or list_categories() to discover valid slugs.' Include a recovery hint.
• Add tool annotations to the FastMCP tool registration. Mark both tools with readOnlyHint=true and idempotentHint=true to signal safety to agents.
• Extend list_businesses to support pagination: add 'offset' (default 0) and 'total_count' to the response. Document: 'Returns up to 'limit' results. Use 'offset' to fetch subsequent pages. Check 'total_count' to know if more results exist.'