Skip to main content
OVR captures screenshots from your UI and diffs them against approved baselines. It runs entirely on your own infrastructure: a web dashboard, a worker that captures and diffs snapshots with Playwright, and a CLI that CI pipelines run to upload a build.

​
Why OVR?

Unit tests verify behavior, not appearance. A component can pass every test and still ship with broken spacing, a clipped label, or a layout that only breaks at one viewport width. That gap matters more as more UI code gets written by AI agents. A generated change can compile, pass its tests, and still be visually wrong in a way that’s easy to miss in a text-based review. Asking an AI to look at a screenshot and judge whether it looks right helps, but that judgment is non-deterministic and rarely covers every breakpoint and state a change could touch. Pixel diffing is deterministic: every story, every configured viewport, every time. It won’t tell you if a design is good, but it will tell you exactly what changed, and let a reviewer, human or automated, decide if that was intentional.

​
How it works

  • No separate test files. Point the CLI at a storybook build output instead of writing and maintaining dedicated visual test files.
  • Review workflow. Visual diffs that cross the threshold go into a review queue.
  • Git status checks. Connect a project to your Git provider and OVR posts build/review status as a commit check, so a pull request can gate on it.
  • Bring your own infra. Postgres, Redis, and any S3-compatible storage. Self-hosted or managed, your call.

​
Source and license

OVR is source-available under the Elastic License 2.0: free to self-host and modify. The one restriction: you may not offer it to third parties as a hosted or managed service.