A production-grade multi-agent pipeline built with LangGraph and Mistral AI.
Three specialized agents collaborate to research, write, and review content — coordinated by a LangGraph state machine.
┌──────────────┐ ┌─────────────┐ ┌──────────────┐
│ Researcher │────▶│ Writer │────▶│ Reviewer │
│ Agent │ │ Agent │ │ Agent │
└──────────────┘ └─────────────┘ └──────────────┘
│ │ │
└───────────────────┴────────────────────┘
│
┌──────▼──────┐
│ Coordinator │
│ (Router) │
└─────────────┘
LangGraph StateGraph
Each agent sets next_agent in the shared state. The Coordinator reads this value and routes to the correct next node using add_conditional_edges.
| Agent | Role | Temperature |
|---|---|---|
| Researcher | Deeply analyzes topic, produces structured notes | 0.3 (factual) |
| Writer | Turns research into engaging blog post | 0.7 (creative) |
| Reviewer | Edits draft, returns critique + polished final | 0.4 (precise) |
git clone https://github.com/YOUR_USERNAME/multi-agent-system
cd multi-agent-system
pip install -r requirements.txtcp .env.example .env
# Edit .env and add your Mistral API keyGet your free key at: https://console.mistral.ai
streamlit run app.pymulti-agent-system/
├── agents/
│ ├── __init__.py
│ ├── researcher.py ← Researcher Agent
│ ├── writer.py ← Writer Agent
│ └── reviewer.py ← Reviewer Agent
├── graph/
│ ├── __init__.py
│ └── workflow.py ← LangGraph StateGraph + Coordinator
├── app.py ← Streamlit UI
├── requirements.txt
├── .env.example
└── README.md
Designed a multi-agent LangGraph pipeline with specialized Researcher, Writer, and Reviewer agents — coordinator dynamically routes task execution using Mistral AI, deployed via Streamlit
- LangGraph — State machine and agent orchestration
- Mistral AI — LLM backbone (
mistral-large-latest) - LangChain — LLM abstraction layer
- Streamlit — Web UI
- Python 3.10+
Add screenshots here after running the app.
MIT