This is a Vue.js frontend application, not an MCP server. The codebase contains API wrapper functions (checkLoginStatus, fetchUserInfo, etc.) but no MCP protocol implementation, tool registration, or server infrastructure. Tools are inferred from function exports in src/api/index.ts and src/api/aibot.ts, not explicitly registered as MCP tools. No input schemas are visible for any tool. Descriptions are bilingual but minimal (10-50 chars). No error handling guidance, no output schemas documented, no parameter validation. The package.json lists Vue, Vite, and GitCoffee UI dependencies, this is a frontend app, not an MCP server.

• Convert this frontend app into an MCP server by implementing the MCP protocol (stdio or HTTP transport). Register tools via the MCP tools capability with proper schemas and descriptions.
• Add JSON Schema input definitions for all 37 tools. At minimum: type, required fields, and descriptions for each parameter. Example: startSeekFlow should declare intervalTime as {type: 'number', minimum: 100, maximum: 60000, description: 'Interval in milliseconds (100-60000)'}.
• Expand tool descriptions to 50-200 characters. Include WHAT the tool does, WHEN to use it, and any prerequisites. Example: 'Fetch the current user's profile information including name, email, and permissions. Call this after login to populate the UI.'
• Document output schemas for all tools. Specify which fields are returned, their types, and which are required for downstream tool calls. Example: fetchUserInfo should return {userId, email, name, permissions: string[]}.
• Add error handling with recovery guidance. Instead of silent failures, return structured errors: {error: 'User not found', suggestion: 'Try checkLoginStatus() first to verify authentication.', retryable: false}.