8 tools with basic schemas and descriptions, but significant gaps. All tools have input schemas with type definitions and descriptions present, meeting minimum viability. However, descriptions are terse (avg 80 chars, well below 194-char baseline), lack WHEN/WHY context, and omit error guidance. No output schemas documented. Parameters lack format constraints and validation hints. Tool names follow verb_noun pattern (good), but descriptions read like API docs, not LLM-optimized guidance. No tool annotations (readOnlyHint, destructiveHint). Error handling absent, no recovery guidance or actionable error messages. Composition is reasonable (single responsibility per tool), but missing chaining IDs in responses (e.g., get_partner_messages should return partner_id for downstream calls). Vietnamese descriptions mixed with English reduce clarity for English-speaking LLMs.

• Document output schemas for all 8 tools. Include field names, types, and descriptions. E.g., list_partners should return {partners: [{id, name, stage, last_seen, message_count}], total_count, next_cursor}.
• Expand descriptions to 100 - 200 chars. Add WHEN/WHY context: 'Call this to retrieve all messages in a conversation thread. Use after list_partners to drill into a specific partner's communication history.' Translate to English.
• Add readOnlyHint: true to all tool schemas (via tool annotations in MCP SDK). This signals to agents that these tools are safe to call without confirmation.
• Implement error handling with actionable messages. E.g., if conversation_id is invalid, return 'Conversation not found. Call list_partners first to get valid conversation IDs.' instead of a 500 error.
• Include chaining IDs in responses. E.g., get_partner_messages should return partner_id, page_id, and conversation_id so agents can call downstream tools without extra lookups.