Three tools with complete schemas and descriptions, but significant gaps in parameter documentation and output schema specification. All tools have action verbs (search_, dataset_, query) and clear descriptions (100-150 chars). However, parameter descriptions lack detail on constraints, formats, and dependencies. No output schema documented. Error handling is minimal. The tools follow a logical discovery chain (search → info → query) but lack guidance on when to use each step.

• Document output schemas for all three tools. Example for search_datasets: 'Returns {datasets: [{dataset_id: string, title: string, theme: string, record_count: number}], total_count: number}'. This enables LLMs to extract dataset_id for downstream calls.
• Add constraint details to ODSQL parameters. For 'where': 'ODSQL filter syntax (e.g., year >= 2020 AND city = "Paris"). Use field names from dataset_info(). Invalid syntax returns a 400 error with the offending clause highlighted.' Similar guidance for 'select', 'group_by', 'order_by'.
• Add dependency hints to descriptions. dataset_info: 'Call this after search_datasets() to learn column names before writing ODSQL filters.' query: 'Requires dataset_id from search_datasets(). Use dataset_info() first to see available fields.'
• Enhance error messages in callTool(). Catch ODSQL parse errors and return: 'Invalid ODSQL in [where|select|group_by|order_by]. Call dataset_info(dataset_id) to see available fields. Error: [API message]'
• Add result limit guidance. search_datasets and query descriptions: 'Results are paginated (default limit=20, max 100). Use offset to iterate. If no results, try broader keywords or check dataset_info() for available fields.'