Remove dead video/data/MCP modules and deprecate legacy routes

GEMINI.md

@@ -98,7 +98,7 @@ npm --workspace=apps/web run type-check
 ## Testing
 
 - **pytest** — `pythonpath = src`, `testpaths = tests`, `asyncio_mode = auto`
-- Coverage target: 90 % minimum enforced via `pytest.ini` on `backend`, `enhanced_video_processor`, and `enterprise_mcp_server`
+- Coverage settings are defined in `pyproject.toml` for `youtube_extension`.
 - Default test video ID: `auJzb1D-fag` — **never** use `dQw4w9WgXcQ` (Rick Roll; causes flaky tests due to age-gating)
 - Use real `tempfile`/`shutil` temp dirs; avoid `pyfakefs`
 

apps/web/src/app/api/__tests__/api-route.test.ts

@@ -0,0 +1,12 @@
+import { describe, expect, it } from 'vitest';
+import { GET } from '@/app/api/route';
+
+describe('GET /api', () => {
+  it('identifies deprecated video and retired pipeline routes', async () => {
+    const response = await GET();
+    const body = await response.json();
+
+    expect(body.endpoints.video).toContain('DEPRECATED');
+    expect(body.endpoints.pipeline).toContain('retired');
+  });
+});

apps/web/src/app/api/route.ts

@@ -20,7 +20,8 @@ export async function GET() {
       endpoints: {
         video_pack: 'POST /api/video/pack - Emit a hashed Video Pack from a YouTube URL',
         transcribe: 'POST /api/transcribe - Extract video transcripts',
-        video: 'POST /api/video - Video analysis only',
+        video: 'POST /api/video [DEPRECATED] - Video analysis only; use /api/video/pack',
+        pipeline: 'Legacy /api/pipeline routes are retired',
         health: 'GET /api - This endpoint',
       },
     });

apps/web/src/app/api/video/route.ts

@@ -1,8 +1,7 @@
 import { NextResponse } from 'next/server';
 /**
- * @deprecated Legacy video register/list endpoint. No in-repo callers;
- * superseded by the Video Pack pipeline (`POST /api/video/pack`). Kept for
- * backward compatibility with external API consumers. Do not add new callers.
+ * @deprecated Legacy video analysis endpoint. Retained for external API
+ * compatibility; use the Video Pack pipeline (`POST /api/video/pack`) instead.
  */
 import { waitUntil } from '@vercel/functions';
 import { publishEvent, EventTypes } from '@/lib/cloudevents';

apps/web/src/app/docs/api/page.tsx

@@ -54,9 +54,12 @@ export default function ApiDocsPage() {
           <h1 className="mt-2 font-heading text-4xl font-black">API Reference</h1>
           <p className="mt-3 max-w-2xl text-ink/60">
             UVAI exposes a small, REST-style surface for transcription, event
-            extraction, pipeline execution, and search. All endpoints accept
-            and return JSON unless noted. Streaming endpoints use Server-Sent
-            Events.
+            extraction, and search. All endpoints accept and return JSON unless noted.
+          </p>
+          <p className="mt-3 max-w-2xl text-sm text-ink/50">
+            The legacy <code>/api/pipeline</code> and <code>/api/pipeline/stream</code>
+            routes are retired. <code>GET</code> and <code>POST /api/video</code> remain
+            for compatibility but are deprecated; use <code>POST /api/video/pack</code>.
           </p>
         </header>
 

apps/web/src/lib/__tests__/auth-paths.test.ts

@@ -40,8 +40,8 @@ describe('auth path policy', () => {
     expect(needsAuthentication('/api/workspace/export')).toBe(false);
   });
 
-  it('keeps the retired /api/pipeline surface non-public', () => {
-    // The legacy pipeline routes were removed (Video Pack pipeline is
+  it('keeps the deprecated, retired /api/pipeline surface non-public', () => {
+    // The legacy pipeline routes remain retired (Video Pack pipeline is
     // canonical). They must not be re-added as public by accident.
     expect(isPublicApiPath('/api/pipeline')).toBe(false);
     expect(isPublicApiPath('/api/pipeline/stream')).toBe(false);
@@ -289,4 +289,3 @@ describe('AI route classification (rate-limit budget)', () => {
     expect(isAiRoute('/api/transcribe/status', 'GET')).toBe(true);
   });
 });
-

config/import_mapping.json

@@ -4,7 +4,6 @@
       "youtube_extension.processors.simple_real_processor"
     ],
     "backend.main": [
-      "agents.markdown_video_processor",
       "backend.enhanced_video_processor",
       "backend.video_processor_factory",
       "backend.real_api_endpoints"
@@ -70,7 +69,6 @@
       "youtube_extension.services.api_cost_monitor"
     ],
     "start_enhanced_backend": [
-      "agents.markdown_video_processor",
       "youtube_extension.main"
     ]
   },
@@ -89,10 +87,6 @@
       "backend.main",
       "backend.video_processor_factory"
     ],
-    "agents.markdown_video_processor": [
-      "backend.main",
-      "start_enhanced_backend"
-    ],
     "backend.real_video_processor": [
       "backend.video_processor_factory"
     ],
@@ -179,7 +173,6 @@
     "factory.monitoring": "factory/monitoring.py",
     "backend.video_processor_factory": "backend/video_processor_factory.py",
     "backend.test_real_pipeline": "backend/test_real_pipeline.py",
-    "enterprise_mcp_server": "enterprise_mcp_server.py",
     "create_import_map": "create_import_map.py",
     "tools.collect_todos": "tools/collect_todos.py",
     "monitoring_dashboard": "monitoring_dashboard.py",
@@ -189,7 +182,6 @@
     "backend.real_video_processor": "backend/real_video_processor.py",
     "backend.main_v2": "backend/main_v2.py",
     "factory.__init__": "factory/__init__.py",
-    "notebooklm_processor": "notebooklm_processor.py",
     "agents.interactive_metadata_extractor": "agents/interactive_metadata_extractor.py",
     "agents.mcp_ecosystem_coordinator": "agents/mcp_ecosystem_coordinator.py",
     "factory.models": "factory/models.py",
@@ -208,7 +200,6 @@
     "intelligence_layer.__init__": "intelligence_layer/__init__.py",
     "agents.gemini_video_master_agent": "agents/gemini_video_master_agent.py",
     "scripts.fetch_candidates": "scripts/fetch_candidates.py",
-    "test_enterprise_mcp": "test_enterprise_mcp.py",
     "deploy_all": "deploy_all.py",
     "server": "server.py",
     "scripts.setup_llama_agent": "scripts/setup_llama_agent.py",
@@ -262,7 +253,6 @@
     "agents.openai_dev_task_manager": "agents/openai_dev_task_manager.py",
     "agents.a2a_framework": "agents/a2a_framework.py",
     "mcp_server": "mcp_server.py",
-    "agents.markdown_video_processor": "agents/markdown_video_processor.py",
     "chart_script": "chart_script.py",
     "backend.enhanced_video_processor": "backend/enhanced_video_processor.py",
     "scripts.vision_analyze": "scripts/vision_analyze.py",

docs/CAPABILITIES_AUDIT.md

@@ -50,7 +50,7 @@ If a streaming chat UI is the goal, either adopt the AI SDK's `useChat` + a back
 
 ## MCP ecosystem
 
-MCP servers exist (`mcp-servers/litert-mcp`, `mcp-servers/shared-state`, `src/youtube_extension/mcp/`) using `@modelcontextprotocol/sdk@^1.26.0`, primarily for NotebookLM video-analysis orchestration. They are **loosely coupled** — no dedicated deployment workflow, not wired into the main pipelines.
+MCP server implementations exist in `mcp-servers/litert-mcp` and `mcp-servers/shared-state`. The unused standalone NotebookLM video-analysis server under `src/youtube_extension/mcp/` has been removed. The remaining servers are **loosely coupled** — no dedicated deployment workflow, not wired into the main pipelines.
 
 ## Suggested priority order