🧠 Crypto LLM Assistant — Multi-Server AI System using MCP, LangChain & Streamlit

🚀 Intelligent Multi-Server Crypto Assistant

This project is a fully modular crypto analysis system built with the Model Context Protocol (MCP), where multiple AI-powered servers collaborate to provide live market data, analytics, and news — all coordinated by a central LLM client.

Users can query naturally (“What is the current price of BTC?”), and the system automatically chooses the right tool/server, executes it, and replies in natural human language.

🏗️ Project Architecture

.
├── client/
│ ├── client_host.py # MCP Client that connects to all servers
│
├── servers/
│ ├── market_data_server.py # Fetches real-time crypto prices and top movers
│ ├── news_server.py # Retrieves crypto-related news from APIs
│ ├── analytics_server.py # Performs technical indicators (RSI, MACD, etc.)
│
├── ui/
│ ├── app.py # Streamlit UI Interface
│
├── .env # API keys and environment variables
├── pyproject.toml # Managed by UV for dependency & script control
├── README.md # You’re here
└── streamlit_app.py # (Optional) Streamlit UI for visual exploration

🧩 Key Technologies & Tools Used

⚙️ Features

✅ Modular architecture — Market, News, and Analytics servers run independently
✅ MCP protocol — Secure, standard communication between client and servers
✅ LangChain integration — Smart prompt handling and reasoning
✅ Dynamic tool execution — LLM decides which server tool to call
✅ Real-time crypto data via CoinGecko API
✅ Natural language responses (no raw JSON shown to users)
✅ Streamlit frontend for clean, interactive user experience
✅ Async handling — concurrent request processing with asyncio
✅ Built and managed entirely with uv (no requirements.txt!)

🧠 How It Works (Conceptual Flow)

1. User Query → “What is the price of BTC?”
2. The client (client_host.py) sends the query to the OpenAI LLM.
3. The LLM analyzes available tools (like get_price, get_crypto_news, etc.) from all MCP servers.
4. The LLM returns a JSON instruction such as:

   {"tool": "get_price", "arguments": {"symbol": "BTC"}}
   

5.The client runs that tool on the correct server (Market Data server). 6.The server fetches live data (from CoinGecko API) and returns JSON. 7.The client passes that response back into the LLM. 8.The LLM summarizes the result in natural language, e.g.:

“The current price of Bitcoin (BTC) is approximately $67,421 USD.”

##🧮 Example Tools Implemented

###🪙 Market Data Server

get_price() get_market_data() get_top_movers()

###📰 News Server

get_crypto_news() get_coin_news(symbol: str)

###📈 Analytics Server

get_rsi(symbol: str) get_macd(symbol: str) get_fear_greed_index()

⚡ Setup Instructions

We use uv as the environment and dependency manager.

1️⃣ Clone the Repo

git clone https://github.com/yourusername/crypto-llm-assistant.git
cd crypto-llm-assistant

2️⃣ Create Virtual Environment

source .venv/bin/activate

3️⃣ Install Dependencies

uv sync

4️⃣ Add Environment Variables

Create a .env file:

OPENAI_API_KEY=your_openai_api_key_here

🚀 Running the System

Run Entire Client (Main Orchestrator)

uv run python client/client_host.py

Run Servers Individually (for Debug)

uv run python servers/market_data_server.py
uv run python servers/news_server.py
uv run python servers/analytics_server.py

🧪 Development Mode

To inspect any server in isolation:

uv run fastmcp dev servers/market_data_server.py

You’ll see logs like:

✅ [Server] market_data_server starting up...

💬 Example Conversations

Enter query: What is the current price of BTC?

The current price of Bitcoin (BTC) is approximately $67,421 USD.

Enter query: Tell me the top movers in the crypto market.

Top gainers today include PUMP (+16.5%), IP (+12.6%), and MYX (+9.7%).

Enter query: Tell me about latest crypto news.

Whales scoop $1.7B in Ethereum as exchange balances hit a 9-year low...

🧠 Crypto LLM Assistant — Multi-Server AI System using MCP, LangChain & Streamlit

🏗️ Project Architecture

🌐 Streamlit Frontend

A lightweight Streamlit dashboard lets you interact visually with the assistant. Run Streamlit App

uv run streamlit run streamlit_app.py

Future Enhancements

-Integrate trading simulation tools -Add sentiment analytics via NLP models -Expand MCP servers to support stock & forex data

🙋‍♂️ Author

Abdullah Shahzad
📧 abdullahhunjra@gmail.com
🔗 LinkedIn
💻 GitHub