Single tool 'read_query' has a basic description but critical gaps in schema completeness, parameter documentation, and error handling. The input schema is visible but lacks proper type annotations for the 'params' array items and missing descriptions for parameters. Output schema is undocumented. Error responses are generic and do not guide recovery. No tool annotations (readOnlyHint, etc.) despite being READ_ONLY. Naming is acceptable but descriptions are minimal (18 chars). This server falls well below production baseline.

• Add explicit description to 'query' parameter: 'SQL SELECT query to execute. Supports parameterized queries with ? placeholders. Must be a SELECT statement only (no INSERT, UPDATE, DELETE, or DDL).'
• Add explicit description to 'params' parameter: 'Array of parameter values to bind to ? placeholders in the query, in order. Each value is converted to a string for binding.'
• Document the output schema in the tool description: 'Returns {status: "success"|"error", data: [array of result rows as objects], message?: string}. Each row is a dict with column names as keys.'
• Enhance error messages with recovery guidance: 'Database connection failed: [specific error]. Verify DB_HOST, DB_USER, DB_PASSWORD, and DB_NAME environment variables are set correctly.' and 'Invalid query: [error detail]. Ensure query is a SELECT statement and parameters match placeholder count.'
• Add readOnlyHint annotation to tool schema to signal to LLM that this tool is safe to call without confirmation.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry