Blog
Building Mars-Colony Live with Grok Build
After KaTeX math publishing on airomatic.ai, we pointed Grok Build at something more red-planet: Mars-Colony — a satirical but technically serious simulation and planning tool for colonizing Mars.
Tone-wise, think The Martian problem-solving, Project Hail Mary systems thinking, and The Hitchhiker’s Guide to the Galaxy footnotes. Engineering vocabulary stays real; the mission log is allowed to joke.
What is Mars-Colony?
Mars-Colony is a modular Python app where you run a foothold outpost over sols (Martian days). Each step advances the colony: resources burn and regenerate, robots work and charge, habitat integrity drifts, crew morale moves, and random crises show up uninvited.
Phase 1 systems include:
- Colony overview & mission planning — name, crew size, difficulty, RNG seed, target sols
- Habitat & infrastructure — pressurized modules, integrity, solar efficiency, power budget
- Life support & resources — oxygen, water (recycling), food, energy
- Robot & automation fleet — surveyors, builders, haulers, maintainers
- Crew / colonists — roles, health, morale, overcrowding
- Event & crisis system — dust storms, equipment failures, supply delays, ice finds, solar flares, and the occasional ration-bar birthday
The interactive surface is Streamlit Mission Control: KPIs, resource charts, tabs for each subsystem, and a mission log full of useful warnings dressed as dry humor.
Why this project?
Airomatic ships practical AI-adjacent software in public. Mars-Colony is a full product-shaped sandbox:
- Clean modular architecture (systems, models, simulation engine, UI)
- A real time loop you can step and observe
- Room to grow — player decisions, Grok-generated event narratives, and illustrated crisis stills via image APIs
It’s also a great live-coding target: you can scaffold structure, wire a sim tick, then watch dust storms wreck your solar arrays in the same session.
Stack
| Piece | Choice |
|---|---|
| Language | Python |
| UI | Streamlit (Mission Control dashboard) |
| Charts | Plotly |
| Sim unit | 1 sol ≈ 1 Martian day (~24.66 Earth hours) |
| Agent | Grok Build |
Why Streamlit over Gradio? Multi-panel ops dashboards, session state, and time-series charts are first-class — closer to Mission Control than a single model-demo form.
How to run it (locally)
cd Mars-Colony
source .venv/bin/activate # or: uv venv && uv pip install -r requirements.txt
streamlit run mars_colony/ui/app.pyOpen the local URL (usually http://localhost:8501). Use the sidebar to launch a mission, advance sols, and read the log.
Headless demo:
python -m mars_colony --cli --sols 30 --difficulty nominalWatch the stream
The live session is embedded above — follow along as we scaffold the project, implement Phase 1 systems, and run the colony for the first time.
Live page — always-on embed of the current featured stream: Watch live.
Previous episode — KaTeX math publishing demo (YouTube).
Subscribe on YouTube for notifications when we go live. Questions or topic ideas? Send a message.