Duplicate Detection
Two methods work together to catch repeat reports before they clutter your backlog.
SnagRelay runs two independent checks on every incoming report.
1. Hash-based (exact match)
Runs on every report, on every plan. Console and network errors are normalized (UUIDs, numbers, and other volatile values stripped out) and hashed. If the hash matches an existing report's fingerprint, it's flagged as a duplicate immediately — this catches the same error occurring repeatedly with no meaningful variation.
2. Semantic (AI-based)
Runs asynchronously after a report is created. SnagRelay compares the report's content against recent reports using vector similarity, with a confidence threshold you can adjust — the default requires a fairly high similarity score, with an extra boost for reports on the same page URL. This catches duplicates that describe the same underlying issue differently (e.g. "checkout button doesn't work" vs. "can't complete purchase on Safari"), which a plain hash match would miss.
What happens to a duplicate
The duplicate report is linked to the original (visible as a Duplicate badge in Logs, with a link to the original) rather than being suppressed entirely — you can still see it happened, just grouped with its match instead of creating tracker noise.
Tuning sensitivity
If you're seeing false positives (unrelated reports marked as duplicates) or false negatives (duplicates slipping through), adjust the confidence threshold from your Project's AI Enhancement settings.