ChatTwelve exhibits significant definition quality gaps. While 11 of 12 tools have descriptions (92%), most are generic and under 100 chars, failing to explain WHEN to use each tool or what distinguishes similar operations. Only 3 tools (send_message, stream_message, get_session) have visible input schemas with parameter descriptions. The remaining 9 tools lack documented input schemas entirely, a critical failure for LLM tool selection. Naming is verb-forward (good), but descriptions lack actionable context. No tool documents output schemas. Error handling and recovery guidance are absent. The server conflates session management, chat operations, and prompt management without clear composition boundaries.

• Add complete input schemas to all 9 tools lacking them. Include parameter types, descriptions, and constraints (e.g., prompt_id must be a non-empty string UUID).
• Expand descriptions to 100 - 200 chars. For create_session: 'Create a new isolated chat session. Call this before send_message. Each session maintains separate conversation history and prompt context. Returns session_id for use in subsequent calls.'
• Document output schemas for all tools. Example: 'Returns {session_id: string, created_at: ISO8601, prompt_id: string, message_count: int}'.
• Add error handling guidance to descriptions. Example for delete_prompt: 'Deletes the prompt permanently. Returns 404 if prompt_id not found. Returns 409 if prompt is currently active, deactivate first via activate_prompt with a different prompt_id.'
• Implement confirmation pattern for delete_prompt. Add a dry_run parameter (default false) or require explicit confirmation via a separate confirm_deletion tool.

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

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