Heurema contributions should make AI-agent workflows more bounded, auditable, secure, or useful.
We prefer contributions that improve:
- clear delivery boundaries;
- verification evidence;
- local-first behavior;
- security and least privilege;
- reproducibility;
- vendor neutrality;
- practical team adoption.
For small fixes, open a PR directly.
For new tools, broad rewrites, or behavior changes, open an issue first and describe:
- the problem;
- the workflow affected;
- the proposed boundary;
- the security impact;
- how the change will be tested;
- any vendor-specific assumptions.
- The change has a clear scope.
- The README/docs are updated if behavior changed.
- Tests or reproducible checks are included where possible.
- New dependencies are justified.
- No secrets, customer data, or private context are included.
- The change does not create hidden vendor lock-in.
- Failure modes and rollback are documented for risky changes.
AI-assisted code is acceptable only when the contributor remains responsible for the result.
If AI materially helped produce the change, please verify:
- generated code is understood by the contributor;
- licenses and dependencies are acceptable;
- tests or manual checks were run;
- security-sensitive changes received extra review.
Be direct, specific, and respectful. We optimize for clear engineering discussion, not performative certainty.