18 tools with consistent verb-noun naming (list_, describe_, create_, sample_, get_, export_, generate_, index_, retrieve_, safe_query_). All tools have descriptions (avg ~80 chars, baseline 194). Input schemas present for all tools with typed parameters. However, descriptions are terse and lack context on WHEN to use each tool, dependencies between tools, and recovery guidance. Output schemas are not documented. Error handling is absent, no guidance on retryability, user-fixable vs fatal errors, or actionable recovery steps. Parameter descriptions are minimal (e.g., 'Optional database scope' lacks format/constraint details). No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite all being READ_ONLY. Missing pagination guidance for list tools.

• Document output schemas for all 18 tools. Include field names, types, and descriptions. Example: list_databases should return {databases: [{name: string, type: string}], total: integer}.
• Expand tool descriptions to 100-200 chars. Add WHEN to use (e.g., 'Call this first to discover available schemas before describing tables'), dependencies, and prerequisites. Example: 'List all schemas in the active datasource. Call this before describe_table to find the correct schema name.'
• Add error handling guidance to descriptions. Example: 'If schema not found, call list_schemas() to see available schemas. If no schemas exist, create_snapshot() first.'
• Annotate all tools with readOnlyHint=true in the tool definition. This signals to agents that these tools are safe to retry and have no side effects.
• Add pagination to list_* tools: include limit (default 20, max 100) and offset parameters. Return total_count in response so agents know if more results exist.