# Multi - Agent

This project implements an advanced sequential thinking process based on the Multi - Agent System (MAS), which is built using the Agno framework and provided services by MCP. Compared with simple state - tracking methods, it uses coordinated specialized agents for in - depth analysis and problem decomposition, evolving from a passive thinking recorder to an active thinking processor.

A multi - agent AI infrastructure based on the Model Context Protocol (MCP), providing a standardized communication layer, supporting vector databases, tool integration, and multi - agent collaboration, including core components such as an inspector dashboard, Qdrant vector database, and Docker network.

A simple MCP server implementation based on WebSocket and Redis, used to demonstrate multi - agent system interaction, supporting HTTP connections, SSE broadcasting, and horizontal scaling.

A multi - agent browser automation framework based on AgentKit, which realizes intelligent web page navigation and task execution through division of labor and collaboration, including four core modules: task planning, navigation control, browser operation and result verification.

00_master_mcp is the core coordinator of the MCP multi - agent system, responsible for managing the connections of all downstream service containers, context routing, security policy enforcement, and complex workflow orchestration, and providing a unified SSE endpoint for external tools to access.