AI代理系统

-MCP-Client-using-LangGraph

LangGraph MCP Client + FastMCP Cloud Expense Tracker Server + Stock Price Tool

This project is a full-stack AI Agent System built using LangGraph, MCP (Model Context Protocol), and FastMCP Cloud.
It shows how an AI agent can call tools — both local and cloud-hosted — to perform real-world tasks like:

• ✔️ Fetching live stock prices
• ✔️ Reading/writing from your FastMCP cloud Expense Tracker server
• ✔️ Running web search
• ✔️ Maintaining persistent conversation memory

---

Watch demo here https://youtu.be/MvDsPgjbRu8

Project Highlights

LangGraph MCP Client

A fully asynchronous agent capable of connecting to multiple MCP servers via:

• MCP over SSE (FastMCP cloud)
• MCP over stdio (local tools)
• Automatic tool loading with client.get_tools()

Integrated tools include:

• GPT-OSS-120B (via Groq)
• DuckDuckGo Search
• Stock Price Tool
• Expense Tracker (FastMCP)

---

FastMCP Cloud-Hosted Server

Your Expense Tracker server is live at: https://Expense-tracker.fastmcp.app/mcp Exposes tools like:

• add_expense
• get_expenses
• delete_expense
• get_total
• etc.

---