Two tools with clear naming and reasonable descriptions, but significant gaps in schema completeness and parameter documentation. list_job_sites has no input schema (empty object) and minimal description. search_jobs has a comprehensive input schema with 15 parameters, but lacks output schema documentation and several parameters lack proper type constraints (e.g., job_type, description_format should be enums). Error handling returns a dict with error/message fields but provides no recovery guidance. No tool annotations (readOnlyHint, destructiveHint) despite both being read-only operations.

• Add explicit enum constraints to job_type (fulltime|parttime|internship|contract) and description_format (markdown|html) in the search_jobs schema. Replace free-form string types with proper enum definitions.
• Document the output schema for search_jobs. Specify that jobs is an array of objects with fields: title, company, location, salary, description, url, date_posted, job_type, etc. Include field types and whether fields are nullable.
• Add input schema to list_job_sites (currently empty {}). Even though it takes no parameters, the schema should be explicit: {"type": "object", "properties": {}, "required": []}.
• Enhance error messages with recovery guidance. When search_jobs fails, return: 'Invalid site_name. Call list_job_sites() to see valid values: linkedin, indeed, zip_recruiter, ...' instead of raw exception text.
• Add readOnlyHint: true to both tool definitions in the @mcp.tool decorator to signal to agents that these are safe, non-destructive operations.