College Scorecard MCP has clear, verb-prefixed tool names (search_schools, get_school) and complete input schemas with type definitions. Descriptions are present but generic, they lack actionable guidance on when to use each tool, dependencies, or error recovery. Parameter descriptions are minimal (e.g., 'School name to search for' lacks format/constraint details). Output schema is not documented, LLMs cannot predict the structure of returned school objects. No error handling guidance; API errors are raw. The _apiKey parameter violates secret-injection best practices by exposing credentials as a tool parameter rather than server-side injection.

• Move _apiKey to server-side environment variable injection. Remove it from tool parameters. Document in the server README how to configure the API key.
• Document the output schema for both tools. Example: 'Returns an object with fields: id (number), name (string), city (string), state (string), tuition_in_state (number|null), admission_rate (number|null), median_earnings_10yr (number|null), completion_rate (number|null), median_debt (number|null).'
• Add error recovery guidance. For 404 errors, suggest 'No school found with that ID. Try search_schools() to find the correct school ID.' For API errors, suggest 'API rate limit or service error. Retry in 30 seconds or check api.data.gov status.'
• Enhance parameter descriptions with constraints. E.g., 'state: Two-letter US state code (e.g., CA, NY, TX). Case-insensitive.' and 'limit: Number of results (1 - 100, default 10).'
• Add tool-level dependency hints. E.g., search_schools description: 'Search for schools by name and state. Returns school IDs needed for get_school(). Use this first to discover schools, then call get_school() for detailed financial and outcome data.'