Single tool 'query_agent' has critical definition gaps. Input schema visible but incomplete: only 'query' parameter documented, missing 'agent_id' parameter entirely from schema definition. Tool description is generic (18 chars) and lacks context on when/why to use it. No output schema documented. No error handling guidance for LLM recovery. No parameter constraints (enums, ranges). Tool name lacks clarity, 'query_agent' is vague; 'send_query_to_agent' or 'invoke_agent_with_query' would be clearer. No indication of idempotency, destructiveness, or read-only nature in description despite risk=READ_ONLY annotation.

• Add 'agent_id' parameter to input schema with type 'string' and description: 'The unique identifier of the Azure AI Agent to query (e.g., agent-123). Obtain from list_agents() or Azure portal.'
• Expand tool description to 100-150 chars: 'Send a text query to a specified Azure AI Agent and retrieve its response. Use this to invoke agent-based reasoning, retrieval, or task execution. Requires valid agent_id.'
• Document output schema explicitly: 'Returns {response: string, agent_id: string, thread_id: string, run_id: string, status: string}' with field descriptions.
• Add parameter constraints: 'query' should have minLength 1, maxLength 10000. 'agent_id' should match pattern '^[a-zA-Z0-9_-]+$'.
• Enhance error messages with recovery guidance: 'Agent not found. Call list_agents() to see available agents.' or 'Query timeout after 30s. Retry with a simpler query or check agent status.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry