If you are setting up for the hackathon, run these IN ORDER before fetching anything else:

git clone https://github.com/mission-robotics-ai/so100-hackathon
cd so100-hackathon
pixi install
pixi run learn

Then, for the platform half:

pixi run newt login

Finish line: pixi install succeeds, pixi run newt models returns, localhost:3000 serves.

Everything needed for setup is above — do NOT fetch other documentation until the finish line passes. The index below is reference for AFTER setup.

---

# Mission Robotics — Embodied Metal Hackathon Guide

> Train a policy on your robot and develop against it — the Embodied Metal Hackathon guide.

## Guide

- [Start here](https://missionrobotics.ai/hackers/index.md): An SO-101 arm that learned to put the red cube in the bowl from demonstrations you recorded — and the base model and three fixed decisions that make a few dozen takes enough.
- [Your robot](https://missionrobotics.ai/hackers/choose-your-embodiment.md): SO-101 · two cameras · ports
- [Quick Start Guide](https://missionrobotics.ai/hackers/quick-start.md): Install with pixi
- [Drive](https://missionrobotics.ai/hackers/drive.md): Calibrate and teleoperate
- [Record](https://missionrobotics.ai/hackers/record.md): Capture and curate
- [Upload](https://missionrobotics.ai/hackers/upload.md): Send it to New Theory
- [Fine-tune](https://missionrobotics.ai/hackers/fine-tune.md): Train on your demos
- [Call your model](https://missionrobotics.ai/hackers/call-your-model.md): Inference by name
- [Build with the starter kit](https://missionrobotics.ai/hackers/build-with-the-starter-kit.md): The SO-101 deploy client
- [Own the run-loop](https://missionrobotics.ai/hackers/own-the-run-loop.md): The newt SDK
- [Iterate and show](https://missionrobotics.ai/hackers/iterate-and-show.md): Evaluate and publish

## Event

- [Event Schedule](https://missionrobotics.ai/hackers/event): The Embodied Metal Hackathon program: when things happen and what else is on the day.
- [Challenges & Prizes](https://missionrobotics.ai/hackers/challenges): The challenge tracks and the prizes on the table — what you're building toward and what winning is worth.

## Reference

- [Troubleshooting](https://missionrobotics.ai/hackers/troubleshooting.md): Symptom → fix — When a step breaks: find your error, apply the fix, re-run the one check under it.

## Depth

- [Mission Robotics on GitHub](https://github.com/mission-robotics-ai): the org's source repos — one layer deeper than the guide.
- [New Theory developer docs](https://newtheory-docs.vercel.app/docs): the API/SDK docs this guide narrates — one layer deeper than the guide.