# Screenshot Diff Baselines

## Objective

Give approved screenshots an owner, a review step, and a rule for updating them.

A stored set of approved reference images with recorded context, a review flow, and rules for updating and retiring them.

## Before You Begin

This feature is being added to an application that already exists and already
works. Do not scaffold a new project, and do not assume a blank slate.

Inspect the codebase first and establish:

- The existing application structure and where code of this kind already lives.
- The framework and version in use.
- The existing design system — colours, spacing, typography, and component conventions.
- Existing UI components you can reuse instead of writing new ones.
- The existing database structure, if this feature needs to persist anything.
- The existing authentication and authorization system, if this feature is user-scoped.
- Dependencies already installed, so you don't add a library that duplicates one.
- The existing test setup and conventions.

Only start writing code once you understand the above. If the application
already implements part of this feature, extend it rather than replacing it.

## Implementation Instructions

1. Store each approved image alongside the exact conditions it was captured under — theme, viewport, pointer type, locale, and the component state — so a reviewer can tell what they are looking at without opening the test.
2. Require a diff to be reviewed image by image. A reviewer who accepts a change must have seen that specific image, and the record must say who approved it and when.
3. Version baselines by platform or browser only where the rendering genuinely differs, because a per-platform copy of every image multiplies the review burden without catching anything new.
4. Delete a baseline when the surface it covers is removed, and report images that no test has requested in recent runs so dead references do not accumulate.
5. Capture, stabilisation, and comparison belong to Visual Regression Testing; this brief owns where approved images live, who signs them off, and when they expire. Do not add capture logic here.

## UI and UX Requirements

Match the application's existing design system exactly. Reuse its components,
spacing, and typography. This feature should look like it was always there.

## Responsive Requirements

Works on mobile, tablet, and desktop. Touch targets are large enough to hit on a
phone, and nothing overflows horizontally at 320px.

## Accessibility Requirements

- Fully keyboard navigable.
- Correct semantic elements and ARIA roles.
- Visible focus states.
- Meets WCAG AA contrast.
- Dynamic changes are announced to screen readers.
- Respects prefers-reduced-motion.

## Edge Cases

- A single approve-all control lets an unrelated regression ride in behind an intentional redesign, which is precisely how a suite stops finding real bugs — accept per image, or require an explicit reason for a bulk accept.
- An image with no recorded theme, viewport, or state is unreviewable, because the reviewer cannot tell whether the difference is a bug or a different configuration entirely.
- Copying every baseline for every operating system triples storage and review time for surfaces that render identically, so per-platform variants must be justified case by case.
- Baselines for deleted components linger and quietly rot, so removal has to be part of the same change that removes the component rather than a periodic cleanup that never happens.
- Approved images stored in the repository inflate its size over time; decide early whether they live in version control or an external store, and make the retrieval work for a fresh checkout.
- Two branches approving different baselines for the same surface will conflict on merge, and the resolution must be a fresh review rather than picking whichever side is newer.
- An approval made against a stale baseline must be invalidated if the underlying test conditions change, or the record claims a sign-off that never happened.

## Testing

Exercise the feature end to end in the running application. Cover every edge case
above, then run the existing test suite and confirm nothing regressed.

## Acceptance Criteria

- [ ] Every stored baseline records its theme, viewport, and component state.
- [ ] Diffs are approved individually, and each approval records an author and a timestamp.
- [ ] Platform-specific baseline variants exist only where rendering genuinely differs, and each is justified.
- [ ] Baselines for removed surfaces are deleted, and unreferenced images are reported.
- [ ] A fresh checkout can retrieve all current baselines without manual steps.
- [ ] Conflicting approvals across branches require a new review rather than resolving automatically.
- [ ] The feature matches the existing design system.
- [ ] No existing functionality is broken.

## Adaptation Rules

- Match the existing design system. Do not introduce a new colour palette,
  spacing scale, or component library.
- Reuse existing components and utilities wherever they fit.
- Follow the naming, file layout, and code style already present.
- Do not upgrade, replace, or remove existing dependencies to make this
  feature fit. Adapt the feature to the app, not the app to the feature.
- Do not break existing functionality. If a change is genuinely required in
  existing code, make the smallest one that works and say so.
- If something in these instructions conflicts with how the application is
  built, follow the application and explain the deviation.

## Final Verification

Before you report the work as done:

1. Re-read the acceptance criteria above and check each one against what you
   actually built.
2. Run the application and exercise the feature end to end.
3. Run the existing test suite and confirm you have broken nothing.
4. Check the feature on mobile, tablet, and desktop widths.
5. Check keyboard navigation and focus handling.
6. Summarize what changed: files added, files modified, and anything you
   deliberately did differently because of how this application is built.

If any acceptance criterion is unmet, fix it before reporting completion.