Corrections

Flag data that needs review.

HVTracker is still early, so these pages are intentionally lightweight. The goal is to validate who wants alerts, data access, and sponsorship before building heavier workflows.

Corrections & appeals — how it works

Any listed project (or anyone else) can dispute anything we publish: a trust score, a grade, a capability field, provenance drift, category, or listing status. This is the process, in full:

  • What counts as evidence. Scores are computed only from public, checkable signals, so disputes are settled the same way: point at the public source (repo, registry, manifest, commit) that contradicts what we show. "Our score should be higher" isn't reviewable; "your provenance check missed our PEP 740 attestation, here's the PyPI link" is.
  • What happens next. We re-check the signal against the source. If we're wrong, the fix ships as a data or detector change and takes effect on the next render — and if the error was in the methodology itself, that's said plainly in the methodology changelog. If we're right, you get the reasoning, not silence.
  • Turnaround. HVTracker is independently run; expect a response within a week, usually sooner.
  • Public appeal. Prefer a public record, or disagree with the outcome? Open a GitHub issue — disputes and their resolutions stay visible there.

Scores are never changed by request — only by evidence. You can also verify any score yourself against its signed credential.

Request a correction

Spotted wrong data on a listing? Send the repo, explain what is wrong, and optionally leave a contact email for follow-up.