docs: add sub-phase development rules and acceptance test structure

AGENTS.md

@@ -105,14 +105,69 @@ test_phase<N>_<module_or_feature>.py
 - Use `tmp_path` fixture for ChromaDB test instances
 - Each test file must have a module-level docstring describing coverage
 
+## SUB-PHASE DEVELOPMENT
+
+**Workflow**: Plan → Implement → Acceptance Test → Commit
+
+### Sub-Phase Plan Template
+
+Each sub-phase plan (stored in `.plans/`) must include:
+1. **Objective** — What this sub-phase delivers
+2. **Acceptance Criteria** — List of behaviors that must work
+3. **Acceptance Tests** — `test_acceptance_<subphase>.py` file(s) with real environment
+4. **Implementation Tasks** — Atomic steps to complete
+
+### Acceptance Testing Rules
+
+**Unit tests** (`test_phase*.py`) — mocked, fast, CI-safe
+**Acceptance tests** (`test_acceptance_*.py`) — real environment, actual LLM/ASR calls
+
+**Acceptance test requirements**:
+- Run against real services (ChromaDB instance, actual LLM API, ASR if applicable)
+- Name format: `test_acceptance_<subphase>_<feature>.py`
+- Location: `backend/app/test/acceptance/`
+- Use `pytest` markers: `@pytest.mark.acceptance` and `@pytest.mark.slow`
+- Each acceptance test file must have docstring describing real environment setup
+- Acceptance tests run manually before sub-phase completion, not in CI
+
+**Example acceptance test**:
+```python
+"""Acceptance test: Phase 1 RAG query with real Qwen LLM.
+
+Prerequisites:
+- ChromaDB running (local or docker)
+- .env configured with valid LLM_BASE_URL and LLM_API_KEY
+- Test documents ingested via /api/v1/ingest
+"""
+import pytest
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_query_with_real_llm():
+    """Query should return bullet-point answer from actual LLM."""
+    # Real HTTP call to LLM provider
+    # Real ChromaDB retrieval
+    pass
+```
+
+**Sub-phase completion checklist**:
+- [ ] All unit tests pass (`pytest app/test/test_phase*.py -v`)
+- [ ] All acceptance tests pass (`pytest app/test/acceptance/ -v -m acceptance`)
+- [ ] Code reviewed (self or peer)
+- [ ] Sub-phase plan marked complete in `.plans/`
+- [ ] Git commit with clear message referencing sub-phase plan
+
 ## COMMANDS
 ```bash
 # Dev
 backend:  uvicorn app.main:app --reload --port 8000
 frontend: npm run dev
 
-# Test
-backend:  cd backend && pytest app/test/ -v
+# Unit tests (mocked, CI-safe)
+backend:  cd backend && pytest app/test/test_phase*.py -v
+
+# Acceptance tests (real LLM/ASR/ChromaDB)
+backend:  cd backend && pytest app/test/acceptance/ -v -m acceptance
 
 # Prod
 docker-compose up -d

backend/app/test/acceptance/test_acceptance_phase1_ingest.py

@@ -0,0 +1,22 @@
+"""Acceptance test: Phase 1 document ingestion with real ChromaDB.
+
+Prerequisites:
+- ChromaDB running with persistent storage
+- Test PDF and TXT files available in test fixtures
+- Embedding model accessible (local or remote)
+"""
+import pytest
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_ingest_pdf_with_real_embedding():
+    """Should ingest PDF and create embeddings in real ChromaDB."""
+    pass  # TODO: implement with real ChromaDB instance
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_ingest_metadata_extraction():
+    """Should extract and store metadata with real embedding pipeline."""
+    pass  # TODO: implement with real metadata extraction service

backend/app/test/acceptance/test_acceptance_phase1_rag_query.py

@@ -0,0 +1,15 @@
+"""Acceptance test: Phase 1 RAG query with real Qwen LLM.
+
+Prerequisites:
+- ChromaDB running (local or docker)
+- .env configured with valid LLM_BASE_URL and LLM_API_KEY
+- Test documents ingested via /api/v1/ingest
+"""
+import pytest
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_query_with_real_llm():
+    """Query should return bullet-point answer from actual LLM."""
+    pass  # TODO: implement with real HTTP call to LLM provider and real ChromaDB retrieval

backend/app/test/acceptance/test_acceptance_phase2_asr.py

@@ -0,0 +1,23 @@
+"""Acceptance test: Phase 2 ASR transcription with real Qwen ASR model.
+
+Prerequisites:
+- ASR model endpoint accessible (local vLLM or cloud)
+- Test audio file available
+- WebSocket server running
+"""
+import pytest
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_asr_transcribe_real_audio():
+    """Should return accurate transcript from real ASR model."""
+    pass  # TODO: implement with real ASR API call
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+@pytest.mark.asyncio
+async def test_ws_asr_streaming():
+    """Should stream audio chunks and receive transcripts via WebSocket."""
+    pass  # TODO: implement with real WebSocket connection

backend/app/test/acceptance/test_acceptance_phase2_video.py

@@ -0,0 +1,22 @@
+"""Acceptance test: Phase 2 video upload with real file storage.
+
+Prerequisites:
+- Backend server running (uvicorn)
+- uploads/ directory writable
+- Test video file < 300MB available
+"""
+import pytest
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_upload_video_mp4():
+    """Should upload MP4 and return playable video URL."""
+    pass  # TODO: implement with real HTTP POST to /api/v1/upload-video
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_upload_size_rejection():
+    """Should reject files over 300MB with proper error."""
+    pass  # TODO: implement with real file upload attempt

backend/app/test/acceptance/test_integration_phase1.py

@@ -0,0 +1,15 @@
+"""Acceptance test: End-to-end Phase 1 — text query → RAG → answer.
+
+Prerequisites:
+- Full backend running (uvicorn)
+- ChromaDB initialized with test documents
+- LLM provider configured and accessible
+"""
+import pytest
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_e2e_phase1_text_query():
+    """Should ingest document and answer question with source attribution."""
+    pass  # TODO: implement full flow: ingest → query → verify bullet answer

backend/app/test/acceptance/test_integration_phase2.py

@@ -0,0 +1,16 @@
+"""Acceptance test: End-to-end Phase 2 — video → ASR → RAG → answer.
+
+Prerequisites:
+- Full backend running (uvicorn)
+- ChromaDB initialized with test documents
+- LLM and ASR providers configured and accessible
+- Test video file with known audio content
+"""
+import pytest
+
+
+@pytest.mark.acceptance
+@pytest.mark.slow
+def test_e2e_phase2_video_query():
+    """Should upload video, transcribe, and answer from transcript."""
+    pass  # TODO: implement full flow: upload → ASR → query → verify answer