Three tools with complete schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. All tools have action verbs (search_, dataset_, query) and non-empty descriptions (avg 156 chars). However, parameter descriptions are sparse or missing context. The query tool accepts free-form ODSQL strings without validation guidance, and no output schemas are documented. Error handling is minimal, API errors are passed through without recovery guidance. No tool annotations (readOnlyHint, etc.) despite all being read-only operations.

• Document output schemas for all three tools. Example: search_datasets returns {datasets: [{dataset_id, title, theme, record_count}], total_count}. This enables LLMs to chain calls (e.g. extract dataset_id from search result, pass to dataset_info).
• Add JSON Schema constraints to numeric parameters: minumum: 1, maximum: 100 for limit; minimum: 0 for offset. Add pattern or enum constraints to ODSQL parameters (where, select, group_by, order_by) with examples of valid syntax.
• Enhance error messages with recovery guidance. E.g. catch fetch errors and return: 'Dataset not found. Try search_datasets(query="<keyword>") to find available datasets.' instead of raw HTTP status.
• Add tool annotations to inputSchema. Set readOnlyHint: true for all three tools to signal to agents that these are safe, non-destructive operations.
• Expand parameter descriptions with format examples. E.g. where: 'ODSQL filter expression, e.g. year >= 2020 AND city = "Nantes". See OpenDataSoft docs for syntax.' This prevents LLM hallucination of invalid ODSQL.