Evolution Simulation (prototype)

Hex-grid agent simulation with genetics, food/starvation, trade/steal, and a semantic language layer. Optional OpenAI calls for trait invention and language adoption; without a key the sim runs in mock mode.

Design notes (broader vision): GeneralIdea.md.

Stack

• Python 3.11, Flask UI (src/web_server.py + canvas frontend)
• Optional: openai when OPENAI_API_KEY is set
• Docker / docker-compose

Quick start (mock AI)

python -m venv venv
# Windows: venv\Scripts\activate
pip install -r requirements.txt
python src/web_server.py

Open http://127.0.0.1:5000

CLI tick loop (no UI):

python src/main.py

Real OpenAI mode

cp .env.example .env
# put OPENAI_API_KEY=sk-... in .env (never commit .env)

Docker:

docker compose up --build

Pass the key via compose env (see docker-compose.yml). UI shows mockup testing vs real testing.

Layout

src/
  agent.py genetics.py grid.py engine.py language.py
  main.py web_server.py
  static/          # hex canvas UI
data/              # runtime agent JSON (gitignored)

Security

API keys must come from the environment only. If a key was ever committed, revoke it in the OpenAI dashboard.