# AI Image Alt Text

## Objective

Draft alternative text for uploaded images so screen readers get something useful.

A suggested alt attribute offered at upload time, editable and confirmable before the image is used.

## 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. Find every place the app accepts an image and currently stores an empty or absent alt attribute, and offer a suggestion at that point rather than in a separate cleanup screen.
2. Ask first whether the image is decorative. A decorative image takes empty alt text, and forcing a description onto it makes the page worse for the people the feature is meant to serve.
3. Pass the surrounding context — the page heading, the nearby caption, the purpose of the field — so the suggestion describes the image's role, not just its contents. Do not repeat an existing caption verbatim; a screen reader user would then hear the same sentence twice.
4. Present the suggestion as prefilled editable text the user can accept, rewrite, or clear, and never overwrite alt text a human already wrote.
5. Alt text is the accessibility attribute and belongs here. Visible captions and the searchable description of an image belong to AI Image Captioning; the two must write to separate fields so neither overwrites the other.

## 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

- Decorative images must be able to take empty alt text as a deliberate choice, recorded as such rather than left as a missing value the tooling keeps flagging.
- The model must not guess at identity, ethnicity, age, health, or any other sensitive trait, and must not assert context the image does not show. Constrain the request to what is visible and strip speculative descriptions from the output.
- Surrounding context should inform the description without being echoed. Alt text that restates the caption adds nothing and lengthens the listening experience.
- Keep suggestions to a short single sentence focused on why the image is there. A paragraph of description is a worse outcome than a terse one.
- Images that carry instructions, warnings, data, or text a user must act on need human review before publication, because a wrong description of a diagram is worse than none.
- An image containing readable text must have that text in the alt attribute or nearby in the page, not be described as an image of text.
- When the model is unavailable, times out, or refuses, the upload must still complete with the alt field empty and flagged, never blocked and never filled with a placeholder.
- Bulk generation across an existing image library needs a cost ceiling, a queue on the app's existing background-job system, and a resumable position so a failure does not restart from the first image.

## 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 image upload path offers a suggested alt text that the user can edit, accept, or clear.
- [ ] Decorative images can be marked as such and store empty alt text without being re-flagged.
- [ ] Suggestions never assert identity or sensitive traits and never restate an existing caption.
- [ ] Human-written alt text is never overwritten by a generated suggestion.
- [ ] Images carrying instructions or data require review before the suggestion is used.
- [ ] A model failure leaves the upload working with the alt field empty and flagged.
- [ ] Alt text is stored in its own field, separate from any visible caption.
- [ ] 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.