Two-column desktop layout (40/60 split, sticky left panel):
- Left: job header with stage badge, interview countdown chip, research
controls (generate/spinner/refresh/retry), and research sections
(talking points, company, leadership, tech, funding, red flags, A11y)
- Right: tabbed panel (JD + match score/keyword gaps, email history,
cover letter) plus locally-persisted call notes via @vueuse/core
- Mobile (≤1023px): single-column, left content first
- Routing guard: redirects to /interviews if no id, job not found, or
wrong status; calls prepStore.fetchFor on mount, clear on unmount
- Check error from POST /research/generate; only start pollTask on success to prevent unresolvable polling intervals
- Surface contacts and fullJob fetch errors in fetchFor; silently ignore research 404 (expected when no research yet)
- Remove redundant type assertions (as Contact[], as TaskStatus, as FullJobDetail)
- Add @internal JSDoc to pollTask
- Remove redundant vi.runAllTimersAsync() after vi.advanceTimersByTimeAsync(3000) in test
Delete useApiFetch.ts wrapper (returned T|null) and update prep.ts and
prep.test.ts to import useApiFetch from useApi.ts directly, destructuring
{ data, error } to match the established pattern used by all other stores.
Adds usePrepStore (Pinia) for interview prep data: parallel fetch of
research brief, contacts, task status, and full job detail; setInterval-
based polling that stops on completion and re-fetches; clear() cancels
the interval and resets all state. Also adds useApiFetch composable
wrapper (returns T|null directly) used by the store.
- Replace lazy import + scripts.db.get_research with inline SQL via _get_db(),
matching the pattern used by research_task_status and get_job_contacts
- Exclude raw_output from SELECT instead of post-fetch pop
- Change HTTPException in generate_research to positional-arg style
- Update test_get_research_found/not_found to patch dev_api._get_db
- _parse_message now prefers text/html over text/plain so digest emails
retain href attribute values needed for link extraction
- Strip <head>, <style>, <script> blocks before storing to remove CSS/JS
garbage while keeping anchor tags intact
- Remove [:4000] truncation — digest emails need full body for URL regex
- Update test: large body should NOT be truncated (assert len == 10_000)
Setup-form Pinia store with per-stage computed lanes, optimistic status
mutation on move, and API-error rollback. Shallow-copies API response
objects on fetch to prevent shared-reference mutation across tests.
Adds GET /api/interviews to fetch all pipeline-stage jobs in one call,
and POST /api/jobs/:id/move to transition a job between kanban statuses
with automatic timestamp stamping (or rejection_stage capture).
