DEVELOPMENT

Automatic Verification Is Now Live

The verification engine described in our last engineering post is now running continuously across the archive, not just as a design. Software entries with declared verification targets are re-checked on a recurring schedule, and the result — verified, a warning, or an entry with nothing yet configured to check — is shown right on the entry’s page.

Results are broken down by platform, so a Windows build failing a check doesn’t get lumped in with a Linux build that passed fine; each shows its own status. When a check is actively in progress, the page shows that too — which file is downloading, which is being hashed — rather than going quiet until it’s done.

An entry that comes back with a problem is re-checked again within about an hour instead of waiting for its next full cycle, so a transient mirror hiccup gets a chance to resolve itself quickly, and a real problem gets flagged again rather than sitting stale for a day.

This is opt-in per entry and still growing in coverage — not every software entry has verification targets configured yet. We’re filling that in gradually, prioritizing entries where an official checksum is actually published somewhere we can point to, rather than approximating one.