labels: prompting, llm, open_source
Eliminate ambiguity as much as possible as early as possible.
- Start with requirements and completion criteria.
- Turn that into a PR/FAQ.
- Systems diagrams (mermaid) to accompany narrative explanations.
- README for ergonomics.
- Check in on anything that was hallucinated you might want to address upstream.
- Generate tests to formalize the desired behaviors and completion criteria.
- Double check that tests cover desired behaviors and aren't overly complicated/mocked
- Generate non test code.