feat: Implement core application structure with new dashboard, settings, and help pages, and enhance opportunities management with persistence and filtering.

docs/phase-3-opportunities-and-workflow.md

@@ -0,0 +1,48 @@
+# Phase 3 — Opportunities Persistence & Workflow
+
+## Goals
+- Persist opportunity search results in Convex.
+- Enable lead management (status, notes, tags).
+- Add filtering, pagination, and basic bulk actions.
+
+## Scope
+- Write opportunities to Convex during search.
+- Read opportunities from Convex in the UI.
+- Implement lead status changes and notes/tags.
+
+## Detailed Tasks
+1. **Persist opportunities**
+   - Update `/api/opportunities` to call `opportunities.upsertBatch` after scoring.
+   - Dedupe by URL + projectId.
+   - Store analysisId + projectId on each opportunity.
+
+2. **Load opportunities from Convex**
+   - Replace local state-only data with Convex query.
+   - Add pagination and “Load more” to avoid giant tables.
+
+3. **Filtering & sorting**
+   - Filters: status, intent, platform, min relevance score.
+   - Sorting: relevance score, createdAt.
+
+4. **Lead workflow actions**
+   - Status change: new → viewed/contacted/responded/converted/ignored.
+   - Add notes + tags; persist via mutation.
+   - Optional: quick bulk action for selected rows.
+
+5. **UI feedback**
+   - Show counts by status.
+   - Empty states for no results or no saved opportunities.
+
+## Dependencies
+- Requires Phase 1 schema and Phase 2 project + analysis persistence.
+
+## Acceptance Criteria
+- Opportunities persist across refresh and sessions.
+- Status/notes/tags are stored and reflected in UI.
+- Filters and pagination are usable.
+
+## Risks
+- Serper or direct Google scraping limits; need clear UX for failed searches.
+
+## Notes
+- Keep raw search results transient; only store scored opportunities.