# Contributing

This is an open methodology project. The playbook improves with every data point.

## Submit a Case Study

The single most valuable contribution: run the protocol on your own research problem and report back.

Open a GitHub Issue with the label `case-study` including:
1. **Domain** — what field?
2. **Research question** — what problem?
3. **Traditional estimate** — how long alone? (be honest)
4. **Force-multiplied time** — how long actually?
5. **Speedup** — traditional / actual
6. **Deliverables** — what did you produce?
7. **What worked** — which prompts/phases were most effective?
8. **What broke** — where did the LLM fail?
9. **Stack used** — which LLM, tools, environment?

## Improve the Prompts

Have a prompt variation that works better for your domain? Fork, add to `prompts/`, open a PR with label `prompt`.

## Report Speedup Data

Open an Issue with label `speedup-data`. Anonymized is fine. Aggregate data strengthens the evidence base.

## Suggest Improvements

Found a gap? Open an Issue with label `enhancement`.

## License

All contributions fall under [CC BY 4.0](LICENSE.txt). By contributing, you agree to this.