Two tools with partial schemas and descriptions. scrape_businesses has a reasonable description (145 chars) and input schema with 4 parameters, but lacks output schema documentation. write_to_github has a good description (156 chars) and 4 input parameters, but also lacks output schema. Both tools have parameter descriptions, but neither tool's output structure is formally documented. The code shows functional implementations but tool definitions appear inferred from code rather than explicitly registered with MCP metadata. No error handling guidance, no parameter constraints (enums), no idempotency hints.

• Document output schemas for both tools. For scrape_businesses, formally declare the returned array structure with field types (name: string, rating: float|null, category: string|null, etc.). For write_to_github, document what the GitHub connector returns (commit SHA, file URL, etc.).
• Add enum constraints to language parameter in scrape_businesses (e.g., 'en', 'fr', 'de', 'es') instead of free-form strings. This prevents LLM hallucination of invalid language codes.
• Expand proxyConfiguration parameter description to explain valid keys and structure: 'Object with keys: useApifyProxy (boolean), apifyProxyGroups (array of strings: RESIDENTIAL, DATACENTER, etc.)'.
• Add error handling guidance to both tool descriptions. E.g., 'If Google Maps returns no results, try a broader query or different language. If the GitHub connector is unavailable, check that the connector_id is valid and the user has authorized GitHub access.'
• Explicitly register tools with MCP server using proper tool definition metadata (name, description, inputSchema with type and description for each parameter, outputSchema). Ensure tools are discoverable via list_tools().

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

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