Negotiation Agent

MVP implementation of the workflow described in negotiation-agent-workflow.html.

What is included

• main.py: Runs a full buyer vs seller negotiation loop
• graph/: State + workflow nodes + loop orchestration
• agents/: BuyerAgent and SellerAgent logic
• tools/: Market, budget, reputation systems
• dashboard/app.py: Streamlit dashboard
• api/server.py: FastAPI API endpoint for programmatic execution
• tests/: Pytest coverage for graph and node behavior

The loop orchestration uses langgraph (StateGraph + conditional edges).

Setup

python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
copy .env.example .env

Add your OPENAI_API_KEY in .env.

Run

python main.py

Run dashboard

streamlit run dashboard/app.py

Run API

uvicorn api.server:app --reload --port 8000

Then call POST /negotiate.

Run tests

pytest -q