Ritsu provides 11 tools with explicit schemas and descriptions. Most tools have clear verb-noun naming (list_agents, ask_agent, create_agent) and documented parameters with types. However, descriptions are inconsistent in depth, some are detailed (ask_agent: 'Send a message to an agent...'), others minimal (list_agents: 'List all enabled agents...'). Parameter descriptions are present but often terse. Output schemas are not documented in the visible code. Error handling guidance is absent. Tool composition is reasonable (separate create/update/reload), but no pagination support visible for list_agents despite potentially large result sets.

• Document output schemas for all tools. Specify what fields are returned (e.g., list_agents returns [{id, name, description, status}]). Use JSON Schema or explicit type descriptions.
• Consolidate or clearly rename duplicate tool pairs. Either merge search_chats/history_search_chats into one tool with a parameter, or rename one to reflect its specific context (e.g., 'search_my_chats' vs 'search_agent_chats').
• Add pagination to list_agents: include limit (default 20, max 100), offset, and total_count in response. Document in description: 'Returns paginated list of agents. Use limit/offset for large deployments.'
• Add error recovery guidance to descriptions. E.g., ask_agent: 'If agent_id not found, call list_agents to see available agents.' view_skill: 'If skill not found, check the manifest in your system context.'
• Replace update_agent's generic 'patch' parameter with explicit fields (name, description, system_prompt, model, provider, etc.) with type constraints. Document which fields are mutable.