xx

docs/plans/2026-01-21-bugfinder-design.md

@@ -0,0 +1,66 @@
+<section name="problem">
+Review Bugfinder feedback in docs/sprints/phase-ongoing/Bugfinder/rpt1-8 to identify root causes across
+Analytics, Home, Meetings, People, Post Meeting, and Tasks pages, then outline ideal fixes that restore
+data visibility, interaction, and persistence.
+</section>
+
+<section name="findings">
+- analytics_reports (branch finding: treat rpt1-rpt3 as equally urgent)
+  - rpt1 Analytics - Meetings: totals and charts depend on meeting.segments; ListMeetings returns
+    include_segments=false and no summary, so word counts and speaker stats remain zero. Root cause in
+    src/noteflow/grpc/mixins/meeting/meeting_mixin.py#L197 and client/src/pages/Analytics.tsx#L75.
+  - rpt2 Analytics - Performance: metrics come only from server psutil snapshots and heuristic latency
+    (src/noteflow/infrastructure/metrics/collector.py, client/src/components/features/analytics/performance-tab.tsx).
+    No client/server toggle exists, and high resource usage likely reflects ongoing post-processing or
+    other server workloads rather than a UI issue.
+  - rpt3 Analytics - Speech: entity map and speech patterns rely on meeting.segments/words in
+    client/src/components/features/analytics/speech-analysis-tab.tsx. With empty segments, entities are
+    blank and speech metrics show zeros; heuristics are placeholders rather than NER output.
+- home_meetings (branch finding: missing data wiring and core interactions)
+  - rpt4 Home Page: MeetingCard renders meeting.segments[0]?.text so all cards show "No transcript
+    available" when ListMeetings omits segments (client/src/components/features/meetings/meeting-card.tsx).
+    Tasks come from meeting.summary.action_items, but ListMeetings omits summary, so tasks disappear on
+    reload (client/src/pages/Home.tsx, client/src/pages/Tasks.tsx). The Ask AI button has no handler
+    (client/src/components/layout/app-sidebar.tsx). Home uses HOME_RECENT_MEETINGS_LIMIT=5 and slices to 6,
+    conflicting with the requested "last 3" (client/src/lib/constants/timing.ts, client/src/pages/Home.tsx).
+    Summary generation does not update meeting titles, and there is no calendar-link signal in meeting
+    payloads to gate that behavior.
+  - rpt5 Meetings: UI calls listMeetings(limit=MEETINGS_PAGE_LIMIT) once with no offset or pagination UI,
+    even though the API returns total_count (client/src/pages/Meetings.tsx).
+- people_post_meeting (branch finding: People list never populating due to data layer)
+  - rpt6 People: aggregateSpeakers uses meeting.segments and preferences-only speaker names
+    (client/src/lib/audio/speaker.ts, client/src/pages/People.tsx). With listMeetings returning empty
+    segments, speaker metrics are all zero. Speaker renames are local only; the renameSpeaker API exists
+    but is not used.
+  - rpt7 Post Meeting: transcript row clicks only trigger playback in Tauri; in web they only toggle
+    selection (client/src/pages/meeting-detail/use-playback.ts). Speaker renaming is hidden behind
+    double-click on SpeakerBadge and is local-only (client/src/components/common/badges/speaker-badge.tsx).
+    Summary tab is concise and relegated to the right panel; no dedicated view for call metrics or
+    background process status. Action items and entities are read-only.
+- tasks_page (branch finding: task list empty)
+  - rpt8 Tasks: tasks are derived from meeting.summary.action_items, but summaries are not included in
+    ListMeetings, so lists are empty after refresh. Completion is stored in preferences only, and the
+    UI supports list view only (client/src/pages/Tasks.tsx). Backend already has TaskModel but no gRPC
+    endpoints, so edits/ownership/kanban are not supported.
+</section>
+
+<section name="recommendation">
+- Selected strategy: build dedicated aggregate endpoints for analytics, speaker stats, and tasks.
+  - Add gRPC endpoints (e.g., GetAnalyticsOverview, ListSpeakerStats, ListTasks) with pagination and
+    time-series responses to replace client-side aggregation of segments/summary fields.
+  - Client Analytics, People, Home, and Tasks pages should consume these aggregates instead of listMeetings.
+- Selected strategy: TaskModel CRUD as source of truth for action items.
+  - On summary persistence, create/update TaskModel rows linked to action_items for provenance.
+  - Add task CRUD endpoints (list/update/status/delete) to support owner, priority, due date, and kanban.
+  - Remove preferences-only completion state; sync status updates through the API.
+- Selected strategy: move the LLM summary into the transcript panel and repurpose the right panel.
+  - Transcript pane shows verbose summary as the primary view; summary generation should target higher
+    verbosity templates by default.
+  - Right panel becomes call metrics (model/tokens/latency) plus processing status (summary/entities/
+    diarization, background jobs) and editing surfaces for tasks/entities.
+- Supporting fixes: wire Ask AI to a route or modal, add Meetings pagination using total_count, expose
+  renameSpeaker in the UI for persistent names, add Analytics client/server toggle for performance metrics,
+  and add calendar-link indicators so summary title updates only apply to non-calendar meetings.
+- Transcript interaction: in web mode, highlight transcript rows on click and surface a desktop-playback
+  message until web playback is implemented.
+</section>

docs/sprints/phase-ongoing/Bugfinder/rpt1/Analytics - Meetings.md

@@ -0,0 +1,9 @@
+![pasted_image_04836a5d-3226-48bf-b3ec-527cd202fad0.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\04836a5d-3226-48bf-b3ec-527cd202fad0.png)
+
+Issues
+
+1.  The header cards for Total Words and Unique Speakers has always been 0 and never changed
+    
+2.  The graphs for Speaker Particiaption and Word Count Trends has never visualized any data
+
+

docs/sprints/phase-ongoing/Bugfinder/rpt2/Analytics - Performance.md

@@ -0,0 +1,10 @@
+![pasted_image_fdffb3dd-d491-4e25-b16d-5c8e3e554d1b.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\fdffb3dd-d491-4e25-b16d-5c8e3e554d1b.png)
+
+Issues
+
+1.  The UI here is actually fine, however this is about 30 mins after my last recording and resources are still nearly fully allocated still - I suspect something to be hanging
+    
+
+Changes
+
+1.  If the server and client are on differing IP addresses or devices, then there should be a switch or a toggle to view the Server or Client performance.

docs/sprints/phase-ongoing/Bugfinder/rpt3/Analytics - Speech.md

@@ -0,0 +1,7 @@
+![pasted_image_9b688a19-5857-40fc-832a-02ce5ca149ef.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\9b688a19-5857-40fc-832a-02ce5ca149ef.png)
+
+Issues
+
+1.  Entity Word Map has always been blank
+    
+2.  Speech patters are either mocked data or incomplete

docs/sprints/phase-ongoing/Bugfinder/rpt4/Home Page.md

@@ -0,0 +1,16 @@
+![pasted_image_88eda805-636e-4f39-8f35-bbb772d66f6b.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\88eda805-636e-4f39-8f35-bbb772d66f6b.png)
+
+Issues
+
+1.  Every meeting says “No Transcript Available” no matter how long ago it finished processing
+    
+2.  Tasks and action items dont seem to be persisted
+    
+3.  Ask AI button does not seem to be working
+    
+
+Change requests
+
+1.  Home screen should only show the last 3 recorded meetings, it currenly shows 5.
+    
+2.  Summarization task should also update the meeting title ONLY if the meeting isn’t connected to a calendar appointments

docs/sprints/phase-ongoing/Bugfinder/rpt5/Meetings.md

@@ -0,0 +1,5 @@
+![pasted_image_89fd9ba7-6eeb-4508-89b3-4e72f8cf5b99.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\89fd9ba7-6eeb-4508-89b3-4e72f8cf5b99.png)
+
+Issues
+
+1.  The past recordings displayed does not seem to be paginated and risks excessive resource consumption

docs/sprints/phase-ongoing/Bugfinder/rpt6/People.md

@@ -0,0 +1,7 @@
+![pasted_image_39c6141e-1ae5-44d8-9900-85fbd6755e99.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\39c6141e-1ae5-44d8-9900-85fbd6755e99.png)
+
+Issues
+
+1.   None of the diararized speakers identified in any of the meetings have ended up here.
+    
+2.  Metrics dont seem to be gathered or stored

docs/sprints/phase-ongoing/Bugfinder/rpt7/Post Meeting.md

@@ -0,0 +1,18 @@
+![pasted_image_43050846-ab72-4fa6-b35f-067f01a67155.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\43050846-ab72-4fa6-b35f-067f01a67155.png)
+
+Issues
+
+1.  Clicking on a transcript line doesn’t seem to do anything
+    
+2.  No way to update speaker names
+    
+
+Changes
+
+1.  The LLM Generated Summary should be significantly more verbose and be what’s displayed  as a primary tab in the current transcript component.
+    
+2.  If the summary is moving to the transcript component, then executive summary should be replaced with call metrics, model/token usage, and any additional processes still running in the background.
+    
+3.  Should be able to edit or delete the tasks and action items, identify owners, criticality,  and deadline.
+    
+    1.  similar capabilities should be extended to extracted entities

docs/sprints/phase-ongoing/Bugfinder/rpt8/Tasks.md

@@ -0,0 +1,12 @@
+![pasted_image_a4e73673-65d1-4e02-8a95-d1929974f342.png](file://C:\Users\PC\AppData\Roaming\CherryStudio\Data\Files\a4e73673-65d1-4e02-8a95-d1929974f342.png)
+
+Issues
+
+1.  Tasks only show up when they are first extracted from a recording but never persisted past restarts.
+    
+
+Changes
+
+1.  Should be able to display in List, Card, and Kanban views
+    
+2.  Tasks should display what meeting they came from and link to that post meeting summary

docs/sprints/phase-ongoing/Bugfinder/strategy-b-implementation.md

@@ -0,0 +1,399 @@
+# Sprint Bugfinder Strategy B: Aggregates + Tasks + Summary Layout
+
+> **Size**: XL | **Owner**: TBD | **Prerequisites**: `docs/plans/2026-01-21-bugfinder-design.md`
+> **Phase**: Ongoing - Bugfinder follow-ups
+
+---
+
+## Open Issues & Prerequisites
+
+> ✅ **Review Date**: 2026-01-21 — Strategy B selected for aggregates, TaskModel CRUD, and summary layout.
+
+### Blocking Issues
+
+| ID | Issue | Status | Resolution |
+|----|-------|--------|------------|
+| **B1** | None | ✅ | N/A |
+
+### Design Gaps to Address
+
+| ID | Gap | Resolution |
+|----|-----|------------|
+| G1 | Cache TTL and invalidation rules for analytics aggregates | Default to 60s TTL; invalidate on summary/segment writes if hook available |
+| G2 | Task de-duplication key definition | Use meeting_id + normalized text; ignore case/punctuation |
+| G3 | Summary verbosity default after layout move | Use existing template defaults, add client toggle in follow-up |
+
+### Prerequisite Verification
+
+| Prerequisite | Status | Notes |
+|--------------|--------|-------|
+| TaskModel exists in DB | ✅ | `src/noteflow/infrastructure/persistence/models/organization/task.py` |
+| Action items persisted in summaries | ✅ | `src/noteflow/infrastructure/persistence/models/core/summary.py` |
+| Meeting list supports project filters | ✅ | `src/noteflow/grpc/mixins/meeting/meeting_mixin.py` |
+
+---
+
+## Validation Status (2026-01-22)
+
+### IMPLEMENTED
+
+| Component | Status | Notes |
+|-----------|--------|-------|
+| Analytics aggregate endpoints | ✅ Implemented | `GetAnalyticsOverview`, `ListSpeakerStats` gRPC methods + repositories |
+| Task CRUD API | ✅ Implemented | Domain entity, service, repository, and gRPC endpoints (`ListTasks`, `UpdateTask`) |
+| Analytics UI | ✅ Implemented | `Analytics.tsx` uses `getAnalyticsOverview` and `listSpeakerStats` |
+| People page | ✅ Implemented | `People.tsx` uses server-backed `listSpeakerStats` endpoint |
+| Tasks page | ✅ Implemented | `Tasks.tsx` uses server-backed `listTasks` and `updateTask` |
+
+### IMPLEMENTED (All Components Complete)
+
+**Summary layout in transcript panel** has been implemented:
+- Summary now appears in a collapsible section above the transcript
+- Right panel tabs reduced to Entities and Notes only
+- Summary supports compact horizontal grid layout
+
+### PARTIALLY IMPLEMENTED
+
+| Component | Status | Notes |
+|-----------|--------|-------|
+| Task storage | ✅ Complete | TaskModel in DB, API implemented, task ingestion from summary action items via `_ingest_tasks_from_summary` in `_generation_mixin.py` |
+
+**Downstream impact**: Meeting Detail page (summary layout move).
+
+---
+
+## Objective
+
+Provide server-backed analytics aggregates and task persistence so Analytics, People, and Tasks pages no
+longer depend on loading meeting segments. Move the LLM summary into the transcript panel and repurpose
+the right panel for metrics/status to align with Bugfinder feedback.
+
+---
+
+## Key Decisions
+
+| Decision | Choice | Rationale |
+|----------|--------|-----------|
+| **Analytics data source** | Dedicated aggregate endpoints with in-memory caching | Scales better than listMeetings; avoids heavy payloads |
+| **Task persistence** | TaskModel CRUD as source of truth | Enables edits, ownership, kanban, and sync across devices |
+| **Summary layout** | Summary becomes primary content inside transcript panel | Matches requested UX priority and keeps metrics in sidebar |
+| **Task dedupe** | Only new, deduped tasks on summary regen | Honors user directive; preserves user-edited tasks |
+
+---
+
+## What Already Exists
+
+| Asset | Location | Implication |
+|-------|----------|-------------|
+| TaskModel schema + relationships | `src/noteflow/infrastructure/persistence/models/organization/task.py` | Reuse existing DB table and FK links |
+| Action items in summaries | `src/noteflow/infrastructure/persistence/models/core/summary.py` | Seed tasks from action items |
+| Summary save flow | `src/noteflow/grpc/mixins/summarization/_generation_mixin.py` | Hook task ingestion after summary persistence |
+| Segment + word timing models | `src/noteflow/infrastructure/persistence/models/core/meeting.py` | Aggregate words/speakers for analytics |
+| Usage aggregate query pattern | `src/noteflow/infrastructure/persistence/repositories/usage_event/_aggregations.py` | Reuse structure for analytics queries |
+| Observability metrics endpoint | `src/noteflow/grpc/mixins/observability.py` | Co-locate performance data in analytics UI |
+
+---
+
+## Scope
+
+| Task | Effort | Notes |
+|------|--------|-------|
+| **Domain Layer** | | |
+| Add Task entity + TaskStatus enum | M | Align with DB statuses: open/done/dismissed |
+| Add analytics DTOs (overview, speaker stats, speech metrics) | M | Plain dataclasses for service outputs |
+| **Infrastructure Layer** | | |
+| Build SqlAlchemyTaskRepository | L | CRUD, list filters (status, project_id(s), meeting_id) |
+| Add analytics aggregation queries | L | Meeting trends, word counts, speaker stats, entity frequencies |
+| Add index migration (tasks.status, tasks.meeting_id, segments.speaker_id) | M | Improve query performance |
+| **Application Layer** | | |
+| TaskService (CRUD + summary ingestion) | L | Create tasks from action items (dedupe) |
+| AnalyticsService with TTL cache | L | Query aggregates, cache by workspace/project/date range |
+| **API Layer** | | |
+| Extend proto with analytics + task messages | L | New gRPC methods and enums |
+| Add AnalyticsMixin + TaskMixin | L | Wire into `NoteFlowServicer` and stubs |
+| Update Rust + TS generated clients | L | Regenerate stubs and adapters |
+| **Client Layer** | | |
+| Analytics page uses aggregate endpoints | M | Replace computeAnalytics with API data |
+| People page uses speaker stats endpoint | M | Remove listMeetings dependency |
+| Tasks page uses TaskModel data | L | Persist completion/status, keep project filters |
+| Meeting detail layout: summary in transcript panel | L | Refactor transcript panel to include summary section |
+
+**Total Effort**: XL (1-2 days)
+
+---
+
+## Domain Model
+
+### Task Entity
+
+```python
+# src/noteflow/domain/entities/task.py
+
+class TaskStatus(Enum):
+    """Task lifecycle status."""
+
+    OPEN = "open"
+    DONE = "done"
+    DISMISSED = "dismissed"
+
+
+@dataclass
+class Task:
+    """User-managed task derived from an action item."""
+
+    id: UUID
+    workspace_id: UUID
+    meeting_id: UUID | None
+    action_item_id: int | None
+    text: str
+    status: TaskStatus = TaskStatus.OPEN
+    assignee_person_id: UUID | None = None
+    due_date: datetime | None = None
+    priority: int = 0
+    completed_at: datetime | None = None
+    created_at: datetime = field(default_factory=utc_now)
+    updated_at: datetime = field(default_factory=utc_now)
+```
+
+**Constraints**:
+- TaskStatus limited to open/done/dismissed (match DB constraint).
+- Task dedupe key = meeting_id + normalized text (lowercase, trim, strip punctuation).
+
+---
+
+## API Schema
+
+### Proto Additions (gRPC)
+
+```protobuf
+enum TaskStatus {
+  TASK_STATUS_UNSPECIFIED = 0;
+  TASK_STATUS_OPEN = 1;
+  TASK_STATUS_DONE = 2;
+  TASK_STATUS_DISMISSED = 3;
+}
+
+message Task {
+  string id = 1;
+  string meeting_id = 2;
+  int32 action_item_id = 3;
+  string text = 4;
+  TaskStatus status = 5;
+  string assignee_person_id = 6;
+  double due_date = 7;
+  int32 priority = 8;
+  double completed_at = 9;
+}
+
+message TaskWithMeeting {
+  Task task = 1;
+  string meeting_title = 2;
+  double meeting_created_at = 3;
+  string project_id = 4;
+}
+
+message ListTasksRequest {
+  repeated TaskStatus statuses = 1;
+  int32 limit = 2;
+  int32 offset = 3;
+  optional string project_id = 4;
+  repeated string project_ids = 5;
+  optional string meeting_id = 6;
+}
+
+message ListTasksResponse {
+  repeated TaskWithMeeting tasks = 1;
+  int32 total_count = 2;
+}
+
+message UpdateTaskRequest {
+  string task_id = 1;
+  string text = 2;
+  TaskStatus status = 3;
+  string assignee_person_id = 4;
+  double due_date = 5;
+  int32 priority = 6;
+}
+
+message UpdateTaskResponse {
+  Task task = 1;
+}
+
+message AnalyticsOverviewRequest {
+  double start_time = 1;
+  double end_time = 2;
+  optional string project_id = 3;
+  repeated string project_ids = 4;
+}
+
+message DailyMeetingStats {
+  string date = 1;
+  int32 meetings = 2;
+  double total_duration = 3;
+  int32 word_count = 4;
+}
+
+message AnalyticsOverviewResponse {
+  repeated DailyMeetingStats daily = 1;
+  int32 total_meetings = 2;
+  double total_duration = 3;
+  int32 total_words = 4;
+  int32 total_segments = 5;
+  int32 speaker_count = 6;
+}
+
+message SpeakerStat {
+  string speaker_id = 1;
+  string display_name = 2;
+  double total_time = 3;
+  int32 segment_count = 4;
+  int32 meeting_count = 5;
+  double avg_confidence = 6;
+}
+
+message ListSpeakerStatsRequest {
+  double start_time = 1;
+  double end_time = 2;
+  optional string project_id = 3;
+  repeated string project_ids = 4;
+}
+
+message ListSpeakerStatsResponse {
+  repeated SpeakerStat speakers = 1;
+}
+
+service NoteFlowService {
+  rpc ListTasks(ListTasksRequest) returns (ListTasksResponse);
+  rpc UpdateTask(UpdateTaskRequest) returns (UpdateTaskResponse);
+  rpc GetAnalyticsOverview(AnalyticsOverviewRequest) returns (AnalyticsOverviewResponse);
+  rpc ListSpeakerStats(ListSpeakerStatsRequest) returns (ListSpeakerStatsResponse);
+}
+```
+
+---
+
+## Migration Strategy
+
+### Phase 1: Schema
+1. Add index on `noteflow.tasks.status` and `noteflow.tasks.meeting_id` for list filters.
+2. Add index on `noteflow.segments.speaker_id` for speaker aggregates.
+
+### Phase 2: Backfill
+1. Optional: backfill tasks from existing summaries by replaying summary action items per meeting.
+
+### Migration Risks
+
+| Risk | Mitigation |
+|------|------------|
+| Large segment tables slow aggregate queries | Use TTL cache + add indexes; consider materialized view later |
+| Task dedupe false positives | Normalize text conservatively; log collisions |
+
+---
+
+## Shared Types & Reuse Notes
+
+- **Aggregation pattern**: reuse SQL aggregation style from `src/noteflow/infrastructure/persistence/repositories/usage_event/_aggregations.py`.
+- **Processing status UI**: reuse `ProcessingStatus` component in meeting detail sidebar.
+- **Task priority badges**: reuse `client/src/components/common/PriorityBadge` and `client/src/types/task.ts`.
+
+---
+
+## UI Components (if applicable)
+
+### Meeting Summary Section
+
+```tsx
+// client/src/pages/meeting-detail/summary-panel.tsx
+
+export function SummaryPanel({ summary, summaryMeta, onGenerateSummary }: SummaryPanelProps) {
+  // Render executive summary, key points, and action items inside transcript panel
+}
+```
+
+---
+
+## Deliverables
+
+### Backend
+
+**Domain Layer**:
+- [x] `src/noteflow/domain/entities/task.py` — Task entity + TaskStatus enum
+
+**Infrastructure Layer**:
+- [x] `src/noteflow/infrastructure/persistence/repositories/task_repo.py` — Task CRUD + list filters
+- [x] `src/noteflow/infrastructure/persistence/repositories/analytics_repo.py` — aggregate queries
+- [x] `src/noteflow/infrastructure/persistence/repositories/__init__.py` — export repositories
+
+**Application Layer**:
+- [x] `src/noteflow/application/services/tasks/service.py` — CRUD + summary ingestion
+- [x] `src/noteflow/application/services/analytics/service.py` — cached aggregates
+
+**API Layer**:
+- [x] `src/noteflow/grpc/proto/noteflow.proto` — new messages + RPCs (TaskStatusProto, TaskProto, DailyMeetingStatsProto, SpeakerStatProto, etc.)
+- [x] `src/noteflow/grpc/proto/noteflow_pb2.pyi` — Python type stubs for new proto types
+- [x] `src/noteflow/grpc/mixins/analytics_mixin.py` — GetAnalyticsOverview + ListSpeakerStats
+- [x] `src/noteflow/grpc/mixins/tasks.py` — ListTasks + UpdateTask
+- [x] `src/noteflow/grpc/mixins/converters/_domain.py` — Task/Analytics proto converters
+- [x] `src/noteflow/grpc/service.py` — register mixins (AnalyticsMixin, TasksMixin)
+
+**Migrations**:
+- [x] `src/noteflow/infrastructure/persistence/migrations/versions/v6w7x8y9z0a1_add_task_analytics_indexes.py` — indexes for tasks.status, tasks.meeting_id, segments.speaker_id
+
+### Client
+
+- [x] `client/src/api/types/features/analytics.ts` — analytics request/response types
+- [x] `client/src/api/types/features/tasks.ts` — task request/response types
+- [x] `client/src/api/interface.ts` — new API methods
+- [x] `client/src/api/adapters/tauri/sections/analytics.ts` — analytics endpoint bridge
+- [x] `client/src/api/adapters/tauri/sections/tasks.ts` — task CRUD bridge
+- [x] `client/src-tauri/src/commands/analytics.rs` — Rust Tauri commands
+- [x] `client/src-tauri/src/commands/tasks.rs` — Rust Tauri commands
+- [x] `client/src-tauri/src/grpc/client/analytics.rs` — Rust gRPC client methods
+- [x] `client/src-tauri/src/grpc/client/tasks.rs` — Rust gRPC client methods
+- [x] `client/src-tauri/src/grpc/types/analytics.rs` — Rust type definitions
+- [x] `client/src-tauri/src/grpc/types/tasks.rs` — Rust type definitions
+- [x] `client/src/pages/Analytics.tsx` — use aggregates (uses getAnalyticsOverview, listSpeakerStats)
+- [x] `client/src/pages/People.tsx` — use speaker stats (uses listSpeakerStats endpoint)
+- [x] `client/src/pages/Tasks.tsx` — use TaskModel data (uses listTasks, updateTask)
+- [x] `client/src/pages/meeting-detail/index.tsx` — summary in collapsible section above transcript
+- [x] `client/src/pages/meeting-detail/summary-panel.tsx` — added compact mode with horizontal grid layout
+
+---
+
+## Test Strategy
+
+### Fixtures to extend or create
+
+- `tests/conftest.py`: add Task fixtures for task lists
+- `tests/infrastructure/`: add repository fixtures for analytics aggregates
+
+### Parameterized tests
+
+- Task status transitions: open/done/dismissed
+- Analytics filters: project_id vs project_ids vs none
+
+### Core test cases
+
+- **Domain**: TaskStatus normalization + completion timestamp logic
+- **Service**: summary ingestion dedupe, analytics cache hits/misses
+- **API**: ListTasks pagination + filters, GetAnalyticsOverview range queries
+- **Integration**: task list after summary regenerate adds only new tasks
+
+---
+
+## Quality Gates
+
+- [x] `pytest tests/application/test_task_service.py` passes (19 tests)
+- [x] `pytest tests/application/test_analytics_service.py` passes (10 tests)
+- [x] `pytest tests/grpc/test_tasks_mixin.py` passes (7 tests)
+- [x] `pytest tests/grpc/test_analytics_mixin.py` passes (6 tests)
+- [ ] `make quality` passes
+- [x] No type suppression comments or loose typing introduced
+
+---
+
+## Post-Sprint
+
+- [ ] Add kanban view for tasks and inline editing
+- [ ] Add NER-based entity analytics (NamedEntityModel aggregates)
+- [ ] Materialized analytics tables if cache misses become expensive