Server has 4 tools with explicit Zod schemas and descriptions, but quality is inconsistent. Tools 1 - 2 (add, calculate) are well-defined with clear schemas and English descriptions. Tools 3 - 4 (langchain_request, generate_blog) have complex schemas but descriptions are in Japanese, limiting LLM comprehension. Parameter descriptions are present but sparse (avg ~40 chars vs baseline 72). No output schemas documented. Error handling is present but generic ('エラー: ...'). Tool naming follows verb_noun pattern but lacks clarity on when to use langchain_request vs generate_blog, both call external APIs but serve different purposes.

• Translate all tool and parameter descriptions to English. Use clear, concise descriptions (50 - 200 chars) that explain WHAT the tool does, WHEN to use it, and what it returns.
• Document output schemas for all tools. Specify the structure of returned content (e.g., 'Returns {content: [{type: "text", text: string}]}'). This enables LLMs to chain tools correctly.
• Remove api_key as a tool parameter in generate_blog. Inject FastAPI credentials server-side via environment variables (FASTAPI_API_KEY). Never expose secrets in tool parameters.
• Enhance error messages with actionable recovery guidance. Instead of 'エラー: HTTP 404', return 'FastAPI endpoint not found. Verify fastapi_base_url is correct and the /api/v1/llm/blog/generate endpoint exists.'
• Clarify the distinction between langchain_request (generic HTTP proxy) and generate_blog (specialized blog endpoint). Consider renaming langchain_request to call_external_api or consolidating if they serve the same purpose.