Build a resilient reading-list app.

Design a reading-list app using this supplied fictional JSON fixture: {"items":[{"id":"r1","title":"Designing a clear form","topic":"Design","minutes":6},{"id":"r2","title":"Checking a data table","topic":"Data","minutes":8},{"id":"r3","title":"Making links accessible","topic":"Web","minutes":5}]}. These titles are exercise records, not published articles; an item detail should show its supplied metadata without a fake article link.

Define a typed data contract and sketch loading, populated, empty, failed and retry states. Let the user filter by topic and mark an item as saved locally. No login or server-side personal data is required.

Submit a document, images, presentation or spreadsheet with the contract, state diagram and wireframes. Review criteria: missing or invalid data is considered, every error has a recovery path and the design preserves context while loading.

Build the reading-list app with a local fixture or a local mock endpoint. Add a controllable delay and failure switch so reviewers can reproduce loading behavior. Keep already loaded items visible during refresh and prevent duplicate requests for the same resource.

Store saved item IDs locally, explain the lifetime of any cache and provide a way to reset practice data. Handle an empty array and malformed responses. Use semantic controls, keyboard focus and readable text. Do not embed secrets or depend on a paid service.

Submit a document, images, presentation or spreadsheet with screenshots of success and recovery states, a source link and run instructions. Review criteria: navigation reuses loaded data, retry works, the interface distinguishes an empty list from a network failure, and saved state behaves as documented.

Demonstrate the app with a delayed response, one failed request, an empty list and a malformed payload. Record the request count when revisiting an already loaded screen and after an explicit refresh. Check that a stale request cannot overwrite newer results after a filter or reset.

Write meaningful automated checks where practical and a manual checklist for visual states. Report timings with the device/browser and delay settings used. Add a short README documenting architecture, cache behavior, known limits and how to reproduce each scenario.

Submit a document, images, presentation or spreadsheet with the test evidence, results and repository link. Review criteria: reproducible scenarios, correct request counts, useful recovery and an honest account of what was measured.