Installation

EntroPy requires Python ≥ 3.9. Install the core package from a checkout or sdist:

pip install entropy-ai                # users: install from PyPI
pip install -e .                       # contributors: editable install from the repo
pip install "entropy-ai[langchain]"    # optional: a framework adapter
pip install "entropy-ai[pdf,jupyter]"  # optional: PDF / Jupyter reports

Core dependencies

Always installed:

Optional extras

From scratch (minimal runnable project)

Create a virtual environment, install EntroPy, and add two small files you can run immediately:

# 1. environment
python -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate
pip install entropy-ai

# 2. agent.py
cat > agent.py <<'PY'
def agent(inp):
    # return the agent output (or an entropy.AgentRun)
    return str(inp)
PY

# 3. dataset.json
cat > dataset.json <<'JSON'
{"cases": [{"input": "hello", "expected": "hello"}]}
JSON

# 4. run it
entropy run --agent agent:agent --dataset dataset.json --trials 100

Or evaluate the same thing in pure Python:

from entropy import Suite, Dataset, Case

def agent(inp):
    return str(inp)

dataset = Dataset([Case(input="hello", expected="hello")])
results = Suite(seed=42).run(agent, dataset, trials=100)
print(results)