Hey @LifeAtlas 👋
Found a non-obvious bug in validate-contributor.yml that caused my PR to sit open while others merged instantly.
The problem:
Auto-merge for level 2/3/4 PRs only triggers when three conditions are all true:
level != '1' AND has_contributor == 'true' AND validate_contributor.outputs.valid == 'true'
The has_contributor flag is only set to true when the PR includes a change to contributors/*.json. This is nowhere documented — not in the README, not in CONTRIBUTING.md, and the workflow posts no
comment when it silently skips auto-merge.
What happens in practice:
A contributor submitting a clean level 3/4 PR (submission files only, no contributor JSON touch) gets:
- Automated score comment ✅
- No merge ❌
- No explanation why ❌
They wait indefinitely while contributors who happen to bundle a contributors/ update get auto-merged immediately. Issue #2 in this repo is the same root cause.
Fix options:
- Add a workflow comment when has_contributor == false explaining what's missing
- Or document in the README that every PR must include a contributors/ update
- Or remove the requirement and auto-merge on valid submission alone
I ran into this on PR #388. Happy to raise a PR for whichever fix you prefer.
Hey @LifeAtlas 👋
Found a non-obvious bug in validate-contributor.yml that caused my PR to sit open while others merged instantly.
The problem:
Auto-merge for level 2/3/4 PRs only triggers when three conditions are all true:
level != '1' AND has_contributor == 'true' AND validate_contributor.outputs.valid == 'true'
The has_contributor flag is only set to true when the PR includes a change to contributors/*.json. This is nowhere documented — not in the README, not in CONTRIBUTING.md, and the workflow posts no
comment when it silently skips auto-merge.
What happens in practice:
A contributor submitting a clean level 3/4 PR (submission files only, no contributor JSON touch) gets:
They wait indefinitely while contributors who happen to bundle a contributors/ update get auto-merged immediately. Issue #2 in this repo is the same root cause.
Fix options:
I ran into this on PR #388. Happy to raise a PR for whichever fix you prefer.