Server defines 10 tools with basic schemas and descriptions, but quality is inconsistent. All tools have descriptions (10-100 chars, mostly adequate), and input schemas are present with type definitions. However, parameter descriptions are sparse or generic, output schemas are undocumented, and error handling lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Naming follows verb_noun convention (yandex_search, yandex_tts) but some names are redundant (yandex_asr vs yandex_asr_short). Parameter validation rules are not documented in descriptions. No pagination support despite search tools returning multiple results. Composition is reasonable, tools are single-purpose, but output chaining is not designed (e.g., search results don't include structured metadata for downstream use).

• Document output schemas for all tools. E.g., yandex_search should declare: returns array of {title: string, url: string, snippet: string}. yandexgpt_chat should declare: returns {text: string, tokens_used: number}.
• Add formal enum constraints to parameters instead of listing values in descriptions. E.g., 'voice' parameter should have enum: ['alena', 'filipp', 'jane', 'omazh', 'zahar', 'ermil', 'marina']. 'language' should use enum for supported codes.
• Enforce numeric bounds in schemas. Add minValue/maxValue for 'limit' (1-50), 'temperature' (0-1), 'speed' (0.1-3.0). Document in parameter descriptions: 'Must be between X and Y.'
• Implement pagination for search tools. Add 'offset' and 'limit' parameters (default limit=10, max=50). Return {results: [...], total: number, has_more: boolean} to enable multi-page queries.
• Add tool annotations. Mark yandex_tts with destructiveHint=false, idempotentHint=true (same input → same audio). Mark all read-only tools with readOnlyHint=true. Implement error recovery: 'API key missing. Set YANDEX_SEARCH_API_KEY environment variable.' instead of generic errors.