This project is a Streamlit-based UI that connects to a Modular Computation Platform (MCP) tool server over stdio. It uses a LangGraph ReAct agent powered by Groq's LLaMA-3.3-70B model to process and compute structured prompts involving arithmetic and logic.
- LangGraph ReAct agent using
llama-3.3-70b-versatilevia Groq - Supports dynamic tool loading from MCP servers
- Math tools (
add,multiply) as an example - Async agent execution with LangChain and MCP adapters
- Clean UI built with Streamlit
- Python 3.9 or higher
- Groq API Key
math_server.py(tool server script)
Create a .env file in the root of the project:
GROQ_API_KEY=your_groq_api_key_hereRun using streamlit run client.py
MIT license