Severity
Every bug gets a severity so the urgent ones stand out at a glance. There are four tiers:
| Tier | Means |
|---|---|
| Critical | A severe, broad failure — the app is down, or a core flow is broken for everyone. |
| High | A serious problem affecting many users or an important flow. |
| Medium | A real bug with limited scope or a workaround. |
| Low | Minor — cosmetic, edge-case, or low-impact. |
Qalibr sets severity while it reads and classifies the report, see Judgment & classification. Severity then feeds the Signal score, so a critical bug rises above a cosmetic one even when both were just reported.
Calibration markers
Severity is a starting point, not a verdict — and Qalibr tells you how much to trust each one with a small marker on the severity label:
| Marker | Meaning |
|---|---|
| (none) | Confident. The severity was read from the report and held up. |
~ | A starting point. Either the report didn’t carry a clear severity signal (so it defaults to Medium), or there wasn’t enough context to pin the tier. Treat it as a hint. |
↑ | Floored on purpose. The report concerns a critical production surface — such as payment, checkout, or authentication — so severity is held at High or above. That floor is a fixed, deterministic rule, not a guess; it raises an under-rated tier and confirms one that was already high enough. |
§ | One of your rules set this severity. |
Δ | You set this severity by hand, overriding the estimate. Your call stands until you change it. |
The markers tell you where Qalibr is confident, where it’s hedging, and where the call was yours rather than ours — so when you approve or reject a bug, you know which ones are worth a second look.
The ↑ floor recognizes a fixed English vocabulary of critical surfaces and
failure modes. A report written in another language can still be rated Critical
or High on its own merits, but it won’t trigger the deterministic floor — so
give a payment or checkout outage a second look rather than relying on the
marker to appear.
Filtering by severity
In the dashboard you can filter the master list by severity — Critical, High, Medium, Low — on its own or combined with a status filter, so you can pull up, say, every critical bug still awaiting review. See The dashboard.