nightly-pr-triage

To complete missing dataset metadata in Harbor registry PRs, inspect docs/overrides.yml and update new dataset stubs with valid categories, titles, repos, and arxiv fields. This ensures CI validation passes and the registry surfaces accurate dataset branding.

PR triage for nightly dataset registry updates is the process of identifying new datasets introduced by automated PRs and verifying required metadata fields. It ensures empty stubs are populated with meaningful values before CI validation blocks the merge.

CI validation fails on Harbor registry PRs with empty dataset stubs because the validation scripts enforce data quality by blocking PRs that lack required metadata. You must populate categories, title, repo, and arxiv fields in docs/overrides.yml to unblock the merge.

To use validate_overrides.py to unblock nightly PR merges, run the validation script after updating docs/overrides.yml with complete dataset metadata. The script checks for missing categories, titles, repos, and arxiv fields, ensuring PRs meet CI data quality requirements before merging.

The best way to automate triage for dataset registry overrides is to run a workflow that inspects nightly PRs, identifies new dataset stubs in docs/overrides.yml, fills in missing categories and metadata, and validates the changes using validate_overrides.py to ensure CI passes.

A limitation of triaging Harbor registry PRs with overrides is that the process relies on manually inspecting docs/overrides.yml and providing accurate external metadata like arxiv links. It cannot generate missing metadata automatically and requires validation scripts to pass before merging.