01
Regression
A shared component fixes one route and breaks another
Compare repeated buttons, forms, dialogs and navigation across adjacent routes after a focused AI edit.
Regression QA after AI-assisted changes
Changed a site in Cursor? Review the fix and adjacent flows.
After a series of Cursor edits, review existing CTAs, shared forms and the mobile version of the deployed site. A re-check compares confirmed findings with the previous report; review your own acceptance criteria and language variants separately.
For teams evolving an existing Next.js, React or SaaS repository with Cursor.
01 / risk map
Where Cursor builds need a real browser
Review these scenarios before sharing. The automated report records observable issues on the pages you select.
02
Leftovers
The old flow is still reachable
Find stale links, duplicate forms, temporary copy and routes that never entered the task context.
03
Console
The UI works while failing in the background
Separate harmless warnings from runtime failures, failed requests and hydration mismatches.
02 / visitor trace
Four steps for your manual review
Use this sequence alongside the report to review behavior that depends on your business rules and data.
Cursor deployment / first visit
Preflight turns “seems fixed” into verifiable before/after evidence and separates new regressions from existing debt.
Open a sample report- 01
Reproduce the original bug
Confirm the acceptance criterion Cursor was meant to close.
- 02
Walk the adjacent flow
Inspect shared components and data on other routes.
- 03
Change viewport and language
Catch edits tailored to only one screen or copy length.
- 04
Compare the re-check
Record what disappeared, remained and appeared as a regression.
launch checklist
Before sharing a Cursor link
- The fixed flow passes after a clean reload
- Shared components remain sound on adjacent routes
- No new runtime or hydration errors appear in the console
- RU/EN and 390/1440 preserve the action and hierarchy
03 / exact questions
Cursor QA, without vague promises
What the run can and cannot prove.
Do you need repository access?
No. We evaluate the user-facing result through the URL. The code stays in your environment.
Does this replace CI tests?
No. CI protects known contracts; Preflight walks the visible product to reveal unexpected user-facing effects.
ship with evidence