Currency MCP server has 4 tools with Turkish names and descriptions. All tools have basic input schemas with type and description fields. However, critical gaps exist: (1) Tool names use Turkish verbs (kur_sorgula, kur_cevir) rather than English action verbs, LLMs trained on English patterns will struggle to infer intent. (2) Descriptions are present but minimal (10-30 chars), below the 50-200 char baseline for LLM optimization. (3) No output schemas documented, LLMs cannot predict response structure for chaining. (4) Error handling returns unstructured error objects without recovery guidance. (5) No parameter constraints (enums, ranges) despite currency codes being a fixed set. (6) No tool annotations (readOnlyHint, idempotentHint) despite all tools being read-only. Average per-tool score: 42.

• Rename all tools to English: kur_sorgula → get_exchange_rate, kur_cevir → convert_currency, kur_listele → list_currencies, kur_gecmis → get_rate_history. Use verb_noun pattern matching LLM training data.
• Expand tool descriptions to 50-200 chars. Include WHAT (action), WHEN (use case), and WHAT IT RETURNS (key fields). Example: 'Convert an amount from one currency to another using current rates. Returns converted amount, exchange rate, and date. Use after get_exchange_rate to confirm the rate.'
• Add output schemas to tool definitions. Document response structure: {basarili: boolean, kur: number, tarih: string, aciklama: string, kaynak_api: string} for get_exchange_rate; {basarili: boolean, sonuc: number, ...} for convert_currency.
• Constrain currency code parameters with enums. Query supported currencies from kur_listele at startup and populate enum list. Prevents invalid codes and guides LLM selection.
• Enhance error responses with recovery guidance. Return {error: string, recovery: string, retryable: boolean} instead of bare 'hata' field. Example: {error: 'Currency TXX not found', recovery: 'Call list_currencies to see valid codes', retryable: false}.