CAIRA

Research framework for benchmarking AI incident response agents under adversarial evidence, measuring calibration, and verifying auditability.

Setup

python -m venv venv
source venv/bin/activate          # or venv\Scripts\activate on Windows
pip install -r requirements.txt
python db/seed.py                 # populate mock_soc.db with sample data
export ANTHROPIC_API_KEY="sk-..." # required for LLM agent decisions

Running

# API server (FastAPI + Uvicorn)
uvicorn api.app:app --reload

# Dashboard (Streamlit) — runs on http://localhost:8501
streamlit run dashboard/app.py

Tests

pytest -q