The quality matrix

Kritik audits surfaces: the independently assessable parts of your product. That might be a web app, an iOS app, an admin back-office, a database contract. You pick them when you install Kritik, and the same criteria are then scored the same way on each one. That's what makes it a matrix and not five unrelated reports.

Each criterion gets a maturity score from 0 to 4. It measures how systematically a concern is handled, not whether the code happens to be fine today. The top score means automation or CI makes regressing impossible. And no score without evidence: every assessment points at something real in the code, pinned to a commit.

Criteria roll up into domains (security, privacy, accessibility, performance, architecture, agent-readiness and more), weighted so the load-bearing criteria count more. Each domain lands on a 0–100 score and a grade from A to E. Your profile also weights the domains themselves, so the overall grade reflects what matters for your product.

The rule that keeps it honest: an open critical finding caps the grade (and an open high one caps it too, a little higher). A domain can score well on maturity and still show a capped grade with an asterisk, because one exploitable hole shouldn't disappear behind a pile of good habits.

Read a cell against the others in its row. Same domain, different surfaces: that's where the matrix earns its keep, showing you which part of the product lags and whether a weakness repeats everywhere or lives in one place.