chore: update client submodule and enhance project management documentation

- Updated client submodule reference to the latest commit.
- Revised project management documentation to reflect changes in project roles, settings, and rule inheritance.
- Enhanced roadmap and sprint documentation with detailed objectives and deliverables for upcoming sprints.
- Improved clarity on project scoping and dependencies in the context of workspace management.

All quality checks pass.

docs/roadmap.md

@@ -141,9 +141,9 @@ Sprint 18 (Projects) ─────┬─────────────
 | **15** | Platform Hardening | M | Phase 4 | ⚠️ Partial | Central error taxonomy, OpenTelemetry instrumentation, usage events |
 | **16** | Identity Foundation | L | Sprint 15 | ✅ Ready | User auth mechanism, workspace enforcement |
 | **17** | Custom OAuth Providers | L | Sprint 16 | ✅ Ready | OIDC discovery, Authentik/Authelia presets |
-| **18** | Projects v1 | M-L | Sprint 16 | ❌ Not started | Project entity, meeting/task grouping |
+| **18** | Projects v1 | L | Sprint 16 | ❌ Not started | Project entity, ProjectRole, rule inheritance, UI |
 | **19** | Artifacts v1 | XL | Sprint 18 | ✅ Ready | Upload + chunking + embedding pipeline |
-| **20** | Artifacts v2 | XL | Sprint 19 | 🚫 Blocked | Google Drive / OneDrive connectors |
+| **20** | Artifacts v2 + RAG | XL | Sprint 19 | 🚫 Blocked | External connectors, Qdrant migration, Q&A |
 | **21** | MCP Configuration | L | Sprint 18 | 🚫 Blocked | Scoped MCP registry (workspace defaults + project overrides) |
 | **22** | Rules v1 | XL | Sprint 16, 21 | 🚫 Blocked | Rules schema, auto-record, templates, outputs |
 | **23** | Analytics | M-L | Sprint 15, 22 | 🚫 Blocked | Conflict detection, usage/evaluations tabs |
@@ -300,30 +300,46 @@ All sync infrastructure implemented and tested (validated 2025-12-29):
 ---
 
 ### Sprint 18: Projects v1
-**Size**: M-L | **Owner**: Backend + Client | **Prerequisites**: Sprint 16
+**Size**: L | **Owner**: Backend + Client | **Prerequisites**: Sprint 16
 **Status**: ❌ NOT IMPLEMENTED
 
-> **Objective**: Introduce a first-class container above meetings/tasks.
+> **Objective**: Introduce Projects as first-class container with roles, settings, and rule inheritance.
+
+#### Key Decisions (2025-12-30)
+
+| Decision | Choice |
+|----------|--------|
+| Ownership | Single Workspace (projects scoped to one workspace) |
+| Roles | Project-level (viewer/editor/admin separate from WorkspaceRole) |
+| Rules | Merge/inherit (projects inherit workspace rules, can override) |
+| Migration | Default project per workspace for unassigned meetings |
 
 #### Missing Components
 
 | Component | Required Location |
 |-----------|------------------|
-| ProjectModel | `persistence/models/project.py` |
-| Project domain entity | `domain/entities/project.py` |
-| Project RPCs | proto messages needed |
-| Project UI | `client/src/pages/Projects.tsx` |
-| Task domain entity | Currently ORM-only, no gRPC API |
+| Project entity + ProjectSettings | `domain/entities/project.py` |
+| ProjectRole enum | `domain/identity/roles.py` |
+| ProjectMembership entity | `domain/identity/entities.py` |
+| ProjectModel + ProjectMembershipModel | `persistence/models/project.py` |
+| ProjectRepository + impl | `ports/` + `repositories/` |
+| ProjectService | `application/services/project_service.py` |
+| Project RPCs (8 endpoints) | proto messages needed |
+| Project UI (sidebar, switcher, settings) | `client/src/components/projects/` |
 
 **⚠️ Blocker**: Sprint 21 (MCP Config) and Sprint 22 (Rules) depend on project scoping.
 
 #### Deliverables
-- `src/noteflow/domain/entities/project.py`
-- `src/noteflow/infrastructure/persistence/models/project.py`
-- Project RPCs in proto
-- `client/src/pages/Projects.tsx`
-- `src/noteflow/domain/identity/scope.py` — Scope lattice (workspace → project → resource)
-- `src/noteflow/grpc/_interceptors/scope.py` — Enforce scope on all RPCs
+- `src/noteflow/domain/entities/project.py` — Project, ProjectSettings, ExportRules, TriggerRules
+- `src/noteflow/domain/identity/roles.py` — ProjectRole enum with permissions
+- `src/noteflow/domain/identity/entities.py` — ProjectMembership
+- `src/noteflow/infrastructure/persistence/models/project.py` — ORM models
+- `src/noteflow/application/services/project_service.py` — Lifecycle + rule merging
+- `src/noteflow/grpc/_mixins/project.py` — 8 RPCs (CRUD + membership)
+- `client/src/components/projects/ProjectSidebar.tsx`
+- `client/src/components/projects/ProjectSwitcher.tsx`
+- `client/src/components/settings/ProjectSettingsPanel.tsx`
+- Alembic migrations for projects, memberships, meeting.project_id
 
 ---
 
@@ -350,11 +366,17 @@ All sync infrastructure implemented and tested (validated 2025-12-29):
 
 ---
 
-### Sprint 20: Artifacts v2
+### Sprint 20: Artifacts v2 + RAG Migration
 **Size**: XL | **Owner**: Backend + Client | **Prerequisites**: Sprint 19
 **Status**: 🚫 BLOCKED
 
-> **Objective**: Sync from external sources (directories, cloud drives).
+> **Objective**: External connectors + migrate from pgvector to Qdrant for RAG.
+
+#### Key Decision (2025-12-30)
+
+| Decision | Choice | Rationale |
+|----------|--------|-----------|
+| RAG Backend | Qdrant | Replace pgvector for better scaling, hybrid search, collection-per-project isolation |
 
 #### Verified Assets
 
@@ -363,12 +385,20 @@ All sync infrastructure implemented and tested (validated 2025-12-29):
 | StartIntegrationSync RPC | `noteflow.proto:73`, `grpc/_mixins/sync.py:47-105` |
 | Google OAuth | `oauth_manager.py` with PKCE |
 | CalendarEventModel | `models/integrations/integration.py:188-255` |
+| SegmentModel.embedding | `models/core/meeting.py:202-205` (migration source) |
 
 **Blocked by**: Sprint 19 (Artifacts v1) must be implemented first.
 
 #### Deliverables
 - `src/noteflow/infrastructure/artifacts/sources/` — Directory, Drive, OneDrive
 - `ArtifactSource` domain entity with sync metadata
+- `src/noteflow/domain/rag/` — Chunk, SearchResult, QAResponse, ports
+- `src/noteflow/infrastructure/rag/` — Qdrant client, embedding providers, chunking, indexing
+- `src/noteflow/application/services/qa_service.py` — Q&A orchestration
+- `SearchProject`, `AskQuestion` RPCs
+- `client/src/components/qa/QAPanel.tsx` — Q&A interface with citations
+- `scripts/migrate_to_qdrant.py` — pgvector migration
+- Docker compose Qdrant service with `rag` profile
 
 ---
 
@@ -650,7 +680,12 @@ Week 7-9:   Sprint 23 + Sprint 24 + Sprint 25 ─ all can parallelize
 |----------|--------|--------|-----------|
 | **Auth Model** | 16 | Local-first | Single user per install, design for multi-user. Reduces Sprint 16 from L → M. |
 | **OIDC Config** | 17 | Discovery only | Require issuer URL with `.well-known`. Covers 90%+ of providers. |
+| **Project Ownership** | 18 | Single Workspace | Projects scoped to one workspace; simpler ACL model. |
+| **Project Roles** | 18 | Project-level | Separate ProjectRole (viewer/editor/admin) from WorkspaceRole. |
+| **Rule Inheritance** | 18 | Merge/inherit | Projects inherit workspace rules, can add/modify with precedence. |
+| **Meeting Migration** | 18 | Default project | Each workspace gets 'General' project for unassigned meetings. |
 | **Artifact Scope** | 19 | Per-workspace | Simpler ACL, easier cross-project sharing. |
+| **RAG Backend** | 20 | Qdrant | Replace pgvector with dedicated vector DB for scaling, hybrid search, project isolation. |
 | **MCP Storage** | 21 | Backend DB | Enables team sharing, consistent behavior across devices. |
 | **Rules Execution** | 22 | Backend-evaluated | Audit trails, single source of truth. Client handles triggers only. |
 | **Offline Mode** | 12 | Cached read-only | Prevents silent divergence; no mock writes. |

docs/sprints/phase-5-evolution/sprint-19-artifacts-v1/README.md

@@ -29,6 +29,8 @@ Get any non-meeting corpus into embeddings and retrievable context. Enable RAG o
 ## Decision: Per-Workspace
 
 > **Decided**: Artifacts scoped per-workspace by default, can be linked to projects. Simpler ACL model, easier cross-project sharing.
+> **Default**: If `project_id` is omitted, use the workspace's active project (default project if none set).
+> **Roles**: Workspace OWNER/ADMIN implies ProjectRole.ADMIN for artifact access.
 
 ---
 
@@ -49,11 +51,13 @@ Get any non-meeting corpus into embeddings and retrievable context. Enable RAG o
 |------|--------|
 | `ArtifactModel` (metadata, chunks) | M |
 | `ArtifactChunkModel` with embedding | M |
+| Project scoping (`project_id`, default routing) | S |
+| Artifact repositories + ports | M |
 | File upload endpoint | M |
 | Text extraction (PDF, DOCX, MD) | L |
 | Chunking pipeline | M |
 | Embedding pipeline | M |
-| `RetrieveContext` RPC | M |
+| RAG retrieval RPC (align with Sprint 20 `SearchProject`) | M |
 | Artifact upload UI | M |
 
 ---
@@ -62,11 +66,47 @@ Get any non-meeting corpus into embeddings and retrievable context. Enable RAG o
 
 - `src/noteflow/domain/entities/artifact.py`
 - `src/noteflow/infrastructure/artifacts/` — Storage, chunking, embedding
-- `RetrieveContext` RPC for RAG queries
+- Artifact repositories + ports (domain + infrastructure)
+- RAG retrieval RPC (align response schema with Sprint 20 SearchProject)
 - `client/src/pages/Artifacts.tsx`
 
 ---
 
+## Test Strategy
+
+### Fixtures to extend or create
+
+- `tests/conftest.py`: add `workspace_id`, `project_id`, `user_id`, and `artifact_storage_dir` fixtures; extend `mock_uow` with `artifacts` + `artifact_chunks` repos.
+- `tests/grpc/conftest.py`: add `mock_artifact_repo` and `mock_artifact_chunk_repo` to keep gRPC tests isolated.
+- New `tests/infrastructure/artifacts/conftest.py`: provide `sample_artifact_file_*` (pdf/docx/md/txt), `artifact_content`, `chunking_service`, and `embedding_provider_stub`.
+- New `tests/fixtures/artifacts/`: add small representative files for extraction (ASCII-safe if possible).
+
+### Parameterized tests
+
+- Extraction per file type: PDF, DOCX, Markdown, TXT.
+- Chunking strategies and sizes: sentence/paragraph, `max_chunk_size`, `overlap`.
+- Project scoping: explicit `project_id` vs default active project.
+- Retrieval limits + filters: `limit`, `chunk_type`, `meeting_id`, `artifact_id`.
+
+### Core test cases (behavior + robustness)
+
+- **Domain**: Artifact creation validates required metadata, workspace/project scoping, and default project routing.
+- **Storage**: Uploaded binary is encrypted at rest; decrypt matches input; tampered ciphertext fails cleanly.
+- **Extraction**: Handles empty content, corrupted files, non-UTF8, and oversized documents with actionable errors.
+- **Chunking**: Deterministic chunk IDs, overlap rules enforced, and stable output ordering.
+- **Embedding**: Embedding called once per chunk batch; dimension mismatches surface as clear errors.
+- **Retrieval RPC**: Project scoping enforced; default project applied when `project_id` omitted; returns ordered results.
+
+---
+
+## Shared Types & Reuse Notes
+
+- Reuse encrypted asset storage patterns from `infrastructure/audio/` for artifact binary storage.
+- Reuse vector embedding dimension constant and chunking utilities across meeting + artifact pipelines.
+- Align RAG response types with Sprint 20 (`SearchResult`, `SearchFilters`) to avoid API churn.
+
+---
+
 ## Post-Sprint
 
 - Artifact versioning