Two tools with schemas present but significant quality gaps. ask_consultant has a detailed schema with enums and descriptions, but descriptions are verbose (500+ chars) and contain implementation details rather than LLM-optimized guidance. check_agents_status has minimal schema (single boolean param). Both tools lack output schema documentation. Parameter descriptions are overly technical and include Russian text mixed with English, reducing clarity for LLM parsing. No error handling guidance or recovery paths documented. Tool composition is reasonable (two distinct concerns), but descriptions do not follow the 50-200 char baseline for LLM optimization.

• Rewrite tool descriptions to 50-200 characters, focusing on WHAT the tool does and WHEN to use it. Example: 'ask_consultant' → 'Get expert analysis on technical questions from a council of specialized AI agents. Use when you need multiple perspectives on architecture, code, or strategy decisions.'
• Document output schemas for both tools. Specify what fields are returned, their types, and which fields are needed for chaining to other tools (e.g., if ask_consultant returns agent_responses, document the structure of each response object).
• Translate all parameter descriptions to English only. Remove Russian text and implementation details. Focus on what the LLM needs to know to use the parameter correctly.
• Add error handling guidance to tool descriptions. Example: 'If OpenRouter API is unavailable, check_agents_status will return a diagnostic message. Retry after 30 seconds or ask the user to verify their API key.'
• Add constraints to 'custom_role_prompt' parameter: max length (e.g., 2000 chars), validation rules (no API keys, no secrets), and examples of valid custom prompts.