Server has 13 tools with basic schemas and descriptions in Russian. All tools have input schemas with type definitions and parameter descriptions, meeting minimum requirements. However, descriptions are minimal (10-80 chars, well below the 50-200 char LLM-optimized baseline), lack context on WHEN to use each tool, and do not explain consequences of write/destructive operations. No output schemas documented. Parameter descriptions are terse and lack format/constraint details. Error handling is absent, no recovery guidance, no actionable error messages. Tool naming follows verb_noun pattern (task_search, task_create) but lacks clarity on distinctions (task_update vs task_create for similar operations). No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Russian descriptions reduce LLM comprehension in English-primary agents.

• Expand all tool descriptions to 50-200 characters. Include WHAT the tool does, WHEN to use it (vs similar tools), and what it returns. Example: 'Search tasks by title or description. Returns matching task IDs, titles, and URLs. Use before task_get to find a task by name.'
• Document output schemas for all tools. Specify which fields are returned (task_id, title, url, created_date, etc.) so agents can chain tools. Example: 'Returns: {task_id: string, title: string, url: string, group_id: string, created_date: ISO8601}'
• Add error handling with recovery guidance. Replace raw errors with actionable messages. Example: 'Task not found. Try task_search() with a partial title, or verify the task_id is numeric.'
• Implement confirmation or dry-run for destructive tools (task_delete, kb2_document_delete). Add a 'confirm' parameter or return a confirmation token that must be passed in a second call.
• Translate all descriptions to English and add tool annotations (readOnlyHint, destructiveHint, idempotentHint) to clarify which tools are safe to retry and which modify state.