Server provides 6 utility tools with basic schemas and descriptions. All tools have input schemas with type definitions and parameter descriptions, meeting minimum requirements. However, descriptions are inconsistent in depth and quality. Most tools lack output schema documentation, error handling guidance, and composition hints. Tool naming is generally clear (verb_noun pattern), but some descriptions are generic or incomplete. No tool annotations (readOnlyHint, destructiveHint) despite all tools being read-only. Parameter descriptions exist but lack constraint details (ranges, formats, enums). Error handling is minimal, tools return generic errors without recovery guidance.

• Document output schemas for all 6 tools. Specify field names, types, and what each field represents. E.g., datetime_info returns {iso_8601: string, unix_timestamp: number, weekday: string, iso_week: number}.
• Add readOnlyHint: true to all tool definitions in the MCP server registration. This signals to agents that these tools are safe to call without side effects.
• Replace search_knowledge stub with either a real implementation or remove it entirely. If keeping it, implement actual vector search logic and document the expected response format.
• Add constraint details to parameter descriptions: timezone → 'IANA timezone name (e.g., America/Sao_Paulo, Europe/London); default UTC'; expression → 'Math expression using +, -, *, /, //, %, **, and functions like sin, cos, sqrt, log'; top_k → 'Number of results (1 - 100, default 3)'.
• Enhance error messages with recovery guidance. E.g., json_extract: 'Invalid JSON: {error}. Verify the input is valid JSON.' or 'Path not found: {path}. Available top-level keys: {keys}.' This helps LLMs self-correct.