Server exposes 6 tools with significant definition gaps. Tool names lack clear action verbs (getAllTools, getAllFuncTools are vague; queryAgent is ambiguous). Most descriptions are generic and under 100 chars, failing to explain WHEN to use each tool or what distinguishes them. Input schemas are present for only 2 of 6 tools (callTool, queryAgent); the other 4 have no visible schema definitions. Parameters lack type constraints and detailed descriptions. No output schemas documented. Error handling is minimal, callTool returns a string '调用工具失败' with no recovery guidance. The server acts as a meta-tool aggregator but does not follow composition patterns: getAllTools and getAllFuncTools do nearly identical work (should be one tool with a format parameter). No idempotency guarantees, no pagination for potentially large result sets, no permission gates on destructive operations.

• Add explicit input schemas to getAllTools, getResources, and getPrompts. Even if they accept no parameters, declare 'properties: {}, required: []' to signal completeness.
• Rename tools to follow verb_noun convention: getAllTools → list_tools, getAllFuncTools → list_tools (with format param), callTool → invoke_tool, queryAgent → chat_with_agent.
• Merge getAllTools and getAllFuncTools into a single list_tools tool with a 'format' parameter (enum: ['mcp', 'openai_function']) to eliminate duplication.
• Document output schemas for all tools. E.g., list_tools returns {tools: [{name: string, description: string, inputSchema: object}], total: number}. queryAgent returns {message: string, role: 'assistant', toolCalls?: [{name: string, args: object}]}.
• Enhance error handling in callTool: catch errors, classify as retryable/user-fixable/fatal, and return structured {error: string, code: string, recovery: string} instead of a plain string. E.g., 'Tool not found. Available tools: [list]. Call list_tools() to discover.'.