# Background Job Dashboard

## Objective

An operator console for what the queues are actually doing.

An internal dashboard showing queued, running, and failed work across every queue, with redacted payloads and a narrow set of safe bulk actions.

## 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. Build one operator view that spans all queues, not a page per queue. Show depth, oldest item age, throughput, and failure counts per queue so a backlog is visible at a glance.
2. Put the whole dashboard behind an explicit operator permission. Queue payloads carry customer data and job arguments frequently contain credentials.
3. Redact payloads before rendering. Maintain a per-job-type allow list of arguments safe to display and hide everything else — an argument is hidden until someone decides it is safe, not the reverse.
4. Distinguish a job's retry attempts from separate enqueues of the same job. Group attempts under one logical item and show the attempt count, or an operator will read a retry storm as real volume.
5. Offer retry, cancel, and discard only for job types explicitly marked as safe for that action, and support bulk operations with a hard cap on the batch size and a confirmation naming the count.
6. Do NOT let the dashboard become the load problem. Poll on a bounded interval, cache aggregate counts, and never scan an entire queue to render a page.
7. The queued/running/stalled/failed state model, per-job user-facing progress, and cancellation semantics are owned by Background Job Progress; read that model here rather than defining a second one.

## 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 queue with hundreds of thousands of entries must paginate over a bounded window; counting the whole queue on every page load will take the backend down.
- A job that finishes between rendering and acting on it must fail the action clearly, not retry something already completed.
- Bulk retry on a failing job type re-creates the same failure at scale. Warn before retrying a batch whose recent success rate is near zero.
- A stalled worker leaves jobs marked running forever. Detect by heartbeat age and label them as stalled rather than showing them as healthy work in progress.
- Discarding a job is unrecoverable. Require confirmation naming the job type and count, and record who did it.
- Payload redaction must apply to nested arguments and to the failure backtrace, which regularly quotes the arguments back.
- If the queue backend is unreachable, the dashboard must say so plainly rather than showing zeroes that read as an idle system.

## 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

- [ ] One view shows depth, age, throughput, and failures for every queue.
- [ ] Access requires an operator permission and every action is recorded with an actor.
- [ ] Payload arguments are hidden unless explicitly allow-listed, including inside nested values and backtraces.
- [ ] Retry attempts are visually distinguished from separate enqueues.
- [ ] Retry, cancel, and discard are available only for job types marked safe, and bulk actions are capped and confirmed.
- [ ] Rendering the dashboard never scans a full queue, and a backend outage is shown as an error, not as zero.
- [ ] 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.