amal66 makes proof part of the proposal
This fork turns testing from an implied expectation into a visible part of contributing.
amal66 has added contributor guidance and a pull-request template that ask a simple but valuable question: how do you know this change works?
The guidance maps the available checks so contributors can choose the right level of assurance:
- Backend checks for core product behaviour.
- Frontend checks for the user-facing interface.
- End-to-end checks for complete user journeys.
- Evaluation checks for the quality of AI-assisted outcomes.
- Optional real-service checks, which stay out of the default test run unless the required services are available.
It also sets a clear standard: a feature or fix should include the smallest useful regression test, while optional-service tests must skip cleanly for contributors without those accounts.
Spotted something wrong? Or know the PR text has fresher detail than the writeup above?