commit a1853bb68a464a9f4977629f1fb2e487e47b2ff3
Author: Agent Zero <agent-zero@intentures.org>
Date:   Sat Mar 7 01:27:25 2026 +0100

    feat: initial Tauri desktop shell for BatchTube v1.0

diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
new file mode 100644
index 0000000..4a7f5a9
--- /dev/null
+++ b/.github/workflows/build.yml
@@ -0,0 +1,93 @@
+name: Build BatchTube Desktop App
+
+on:
+  push:
+    tags:
+      - 'v*'
+  workflow_dispatch:
+
+jobs:
+  build:
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - platform: 'macos-latest'
+            args: '--target aarch64-apple-darwin'
+          - platform: 'macos-latest'
+            args: '--target x86_64-apple-darwin'
+          - platform: 'ubuntu-22.04'
+            args: ''
+          - platform: 'windows-latest'
+            args: ''
+
+    runs-on: ${{ matrix.platform }}
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Setup Python 3.11
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Install Python dependencies
+        run: |
+          pip install -r backend/requirements.txt
+          pip install pyinstaller
+
+      - name: Bundle Python backend with PyInstaller
+        run: |
+          cd backend
+          pyinstaller batchtube-server.spec
+          mkdir -p ../src-tauri/binaries
+        shell: bash
+
+      - name: Copy sidecar binary (Windows)
+        if: matrix.platform == 'windows-latest'
+        run: cp backend/dist/batchtube-server.exe src-tauri/binaries/batchtube-server-x86_64-pc-windows-msvc.exe
+        shell: bash
+
+      - name: Copy sidecar binary (macOS arm)
+        if: matrix.platform == 'macos-latest' && contains(matrix.args, 'aarch64')
+        run: cp backend/dist/batchtube-server src-tauri/binaries/batchtube-server-aarch64-apple-darwin
+
+      - name: Copy sidecar binary (macOS x86)
+        if: matrix.platform == 'macos-latest' && contains(matrix.args, 'x86_64')
+        run: cp backend/dist/batchtube-server src-tauri/binaries/batchtube-server-x86_64-apple-darwin
+
+      - name: Copy sidecar binary (Linux)
+        if: matrix.platform == 'ubuntu-22.04'
+        run: cp backend/dist/batchtube-server src-tauri/binaries/batchtube-server-x86_64-unknown-linux-gnu
+
+      - name: Install Linux dependencies
+        if: matrix.platform == 'ubuntu-22.04'
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y libwebkit2gtk-4.1-dev libappindicator3-dev librsvg2-dev patchelf
+
+      - name: Install Rust
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          targets: ${{ matrix.platform == 'macos-latest' && 'aarch64-apple-darwin,x86_64-apple-darwin' || '' }}
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '22'
+
+      - name: Install npm dependencies
+        run: npm install
+
+      - name: Build Tauri app
+        uses: tauri-apps/tauri-action@v0
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          TAURI_SIGNING_PRIVATE_KEY: ${{ secrets.TAURI_SIGNING_PRIVATE_KEY }}
+          TAURI_SIGNING_PRIVATE_KEY_PASSWORD: ${{ secrets.TAURI_SIGNING_PRIVATE_KEY_PASSWORD }}
+        with:
+          tagName: ${{ github.ref_name }}
+          releaseName: 'BatchTube v__VERSION__'
+          releaseBody: 'See CHANGELOG for details'
+          releaseDraft: true
+          prerelease: false
+          args: ${{ matrix.args }}
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..71fbd78
--- /dev/null
+++ b/README.md
@@ -0,0 +1,84 @@
+# BatchTube — Tauri Desktop App
+
+## Architecture
+
+```
+batchtube-tauri/
+├── src-tauri/          # Rust/Tauri shell
+│   ├── src/main.rs     # Spawns Python backend, manages window
+│   ├── Cargo.toml
+│   ├── tauri.conf.json
+│   └── icons/          # App icons
+├── backend/            # Existing Python FastAPI app (unchanged)
+│   ├── api_server.py
+│   ├── worker.py
+│   ├── database.py
+│   ├── extractors.py
+│   ├── config.py
+│   ├── web/            # Frontend HTML/CSS/JS
+│   └── batchtube-server.spec  # PyInstaller spec
+├── .github/workflows/  # Cross-platform CI/CD
+│   └── build.yml
+├── package.json
+├── rollback.sh         # Revert to NSIS installer if needed
+└── README.md
+```
+
+## How It Works
+
+1. Tauri starts → spawns `batchtube-server` sidecar (PyInstaller bundle)
+2. Rust polls `http://localhost:8765/api/health` every 500ms
+3. When server ready → shows webview window loading `http://localhost:8765`
+4. User sees the existing BatchTube web UI inside a native window
+5. On close → Tauri kills the sidecar process
+
+## Build
+
+### Prerequisites
+- Rust (stable)
+- Node.js 22+
+- Python 3.11+
+- PyInstaller: `pip install pyinstaller`
+
+### Development
+```bash
+# Start Python backend first
+cd backend && python api_server.py &
+# Then start Tauri dev mode
+npm run dev
+```
+
+### Production Build
+```bash
+# Bundle Python backend
+cd backend && pyinstaller batchtube-server.spec
+# Build Tauri app
+npm run build
+```
+
+### Cross-Platform (CI/CD)
+Push a tag `v1.x.x` to trigger GitHub Actions build for:
+- Windows (.exe + .msi)
+- macOS (.dmg — Intel + Apple Silicon)
+- Linux (.AppImage + .deb)
+
+## Rollback Plan
+
+If Tauri build fails at any point:
+```bash
+bash rollback.sh
+```
+This reverts to the `v1.0-pre-tauri` git tag.
+The original NSIS Windows installer remains at `BatchTube App/BatchTube-Windows-Build.zip`.
+
+## Pricing Integration
+
+License validation happens at app startup via `http://localhost:8765/api/license`.
+ThriveCart webhook at `https://n8n.intentures.org/webhook/thrivecart` handles purchase events.
+
+| Tier | Price | Limit |
+|------|-------|-------|
+| Free | $0 | 5 videos/batch |
+| Creator | $9/mo | 100/day, 1080p |
+| Pro | $19/mo | Unlimited, 4K |
+| Lifetime | $97 | Unlimited, 4K |
diff --git a/backend/api_server.py b/backend/api_server.py
new file mode 100644
index 0000000..64f02aa
--- /dev/null
+++ b/backend/api_server.py
@@ -0,0 +1,443 @@
+"""
+api_server.py - FastAPI backend for YouTube SIPOC Tool
+Serves the Web GUI, REST API, and Server-Sent Events (SSE) for real-time progress.
+"""
+import sys
+import os
+from pathlib import Path
+
+# Ensure project directory is always on sys.path (needed for some Python installations)
+_PROJECT_DIR = Path(__file__).parent.resolve()
+if str(_PROJECT_DIR) not in sys.path:
+    sys.path.insert(0, str(_PROJECT_DIR))
+
+import json
+import asyncio
+import logging
+from typing import Optional, List
+
+from fastapi import FastAPI, HTTPException, BackgroundTasks, Request
+from fastapi.responses import HTMLResponse, FileResponse, StreamingResponse, JSONResponse
+from fastapi.staticfiles import StaticFiles
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
+import uvicorn
+
+import database as db
+import worker
+import config
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s [%(levelname)s] %(name)s: %(message)s",
+)
+logger = logging.getLogger("api_server")
+
+app = FastAPI(title="YouTube SIPOC Tool", version="1.0.0")
+
+# CORS - allow localhost GUI access
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# Serve static web files
+WEB_DIR = Path(__file__).parent / "web"
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Request / Response Models
+# ─────────────────────────────────────────────────────────────────────────────
+
+class CreateJobRequest(BaseModel):
+    input_type: str                    # url | playlist | csv | handle
+    input_value: str                   # the URL/handle/CSV text
+    output_folder: str                 # absolute path to save files
+    selections: List[str]              # artifact types to extract
+    resolution: str = "best"          # for MP4
+    delay_min: float = 2.0
+    delay_max: float = 7.0
+    openrouter_api_key: str = ""
+    openrouter_model: str = "openai/gpt-4o-mini"
+    summary_prompt: str = ""
+
+
+class ResumeJobRequest(BaseModel):
+    openrouter_api_key: str = ""
+
+
+class BrowseFolderRequest(BaseModel):
+    path: str = ""
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Web GUI Serving
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.get("/", response_class=HTMLResponse)
+async def serve_gui():
+    index_path = WEB_DIR / "index.html"
+    if not index_path.exists():
+        return HTMLResponse("<h1>Web GUI not found. Check the web/ directory.</h1>", status_code=404)
+    return HTMLResponse(index_path.read_text(encoding="utf-8"))
+
+
+@app.get("/app.js")
+async def serve_js():
+    return FileResponse(WEB_DIR / "app.js", media_type="application/javascript")
+
+
+@app.get("/style.css")
+async def serve_css():
+    return FileResponse(WEB_DIR / "style.css", media_type="text/css")
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Job Management API
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.post("/api/jobs")
+async def create_job(req: CreateJobRequest):
+    """Create and immediately launch a new extraction job."""
+    # Validate output folder
+    output_folder = req.output_folder.strip()
+    if not output_folder:
+        raise HTTPException(status_code=400, detail="output_folder is required")
+
+    # Validate selections
+    valid_types = set(config.ARTIFACT_TYPES)
+    invalid = [s for s in req.selections if s not in valid_types]
+    if invalid:
+        raise HTTPException(status_code=400, detail=f"Invalid artifact types: {invalid}")
+    if not req.selections:
+        raise HTTPException(status_code=400, detail="At least one artifact type must be selected")
+
+    # Validate input
+    if not req.input_value.strip():
+        raise HTTPException(status_code=400, detail="input_value is required")
+
+    # Create the folder if it doesn't exist
+    try:
+        Path(output_folder).mkdir(parents=True, exist_ok=True)
+    except Exception as e:
+        raise HTTPException(status_code=400, detail=f"Cannot create output folder: {e}")
+
+    # Create job in DB
+    job_id = db.create_job(
+        output_folder=output_folder,
+        selections=req.selections,
+        resolution=req.resolution,
+        delay_min=req.delay_min,
+        delay_max=req.delay_max,
+        input_type=req.input_type,
+        input_value=req.input_value,
+        summary_prompt=req.summary_prompt or config.SUMMARY_SYSTEM_PROMPT,
+        openrouter_model=req.openrouter_model or config.OPENROUTER_MODEL,
+    )
+
+    # Launch background worker thread
+    worker.launch_job(job_id, openrouter_api_key=req.openrouter_api_key)
+
+    logger.info(f"Created and launched job {job_id}")
+    return {"job_id": job_id, "status": "running"}
+
+
+@app.get("/api/jobs")
+async def list_jobs():
+    """List all jobs (most recent first)."""
+    jobs = db.list_jobs()
+    result = []
+    for j in jobs:
+        stats = db.get_job_stats(j["job_id"])
+        j.update(stats)
+        result.append(j)
+    return result
+
+
+@app.get("/api/jobs/{job_id}")
+async def get_job(job_id: str):
+    """Get job details + stats."""
+    job = db.get_job(job_id)
+    if not job:
+        raise HTTPException(status_code=404, detail="Job not found")
+    stats = db.get_job_stats(job_id)
+    job.update(stats)
+    return job
+
+
+@app.post("/api/jobs/{job_id}/pause")
+async def pause_job(job_id: str):
+    job = db.get_job(job_id)
+    if not job:
+        raise HTTPException(status_code=404, detail="Job not found")
+    if job["status"] not in ("running",):
+        raise HTTPException(status_code=400, detail=f"Cannot pause a job with status: {job['status']}")
+    worker.pause_job(job_id)
+    return {"job_id": job_id, "status": "paused"}
+
+
+@app.post("/api/jobs/{job_id}/resume")
+async def resume_job(job_id: str, req: ResumeJobRequest = ResumeJobRequest()):
+    job = db.get_job(job_id)
+    if not job:
+        raise HTTPException(status_code=404, detail="Job not found")
+
+    if job["status"] == "paused":
+        # Job thread is still alive, just unpause it
+        worker.resume_job(job_id)
+        return {"job_id": job_id, "status": "running"}
+
+    elif job["status"] in ("stopped", "failed", "done_with_errors"):
+        # Re-launch the job from where it left off
+        db.update_job_status(job_id, "pending")
+        worker.launch_job(job_id, openrouter_api_key=req.openrouter_api_key)
+        return {"job_id": job_id, "status": "running", "note": "Job re-launched to resume from last checkpoint"}
+
+    else:
+        raise HTTPException(status_code=400, detail=f"Cannot resume job with status: {job['status']}")
+
+
+@app.post("/api/jobs/{job_id}/stop")
+async def stop_job(job_id: str):
+    job = db.get_job(job_id)
+    if not job:
+        raise HTTPException(status_code=404, detail="Job not found")
+    worker.stop_job(job_id)
+    return {"job_id": job_id, "status": "stopped"}
+
+
+@app.post("/api/jobs/{job_id}/skip")
+async def skip_video(job_id: str):
+    """Skip the currently processing video."""
+    job = db.get_job(job_id)
+    if not job:
+        raise HTTPException(status_code=404, detail="Job not found")
+    worker.skip_video(job_id)
+    return {"job_id": job_id, "action": "skip_sent"}
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Videos & Artifacts
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.get("/api/jobs/{job_id}/videos")
+async def get_videos(job_id: str):
+    videos = db.get_videos(job_id)
+    if not videos and not db.get_job(job_id):
+        raise HTTPException(status_code=404, detail="Job not found")
+    # Enrich with artifact statuses
+    result = []
+    for v in videos:
+        arts = db.get_artifacts_for_video(job_id, v["video_id"])
+        v["artifacts"] = arts
+        result.append(v)
+    return result
+
+
+@app.get("/api/jobs/{job_id}/artifacts")
+async def get_artifacts(job_id: str):
+    return db.get_all_artifacts(job_id)
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Logging
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.get("/api/jobs/{job_id}/logs")
+async def get_logs(job_id: str, limit: int = 500):
+    if not db.get_job(job_id):
+        raise HTTPException(status_code=404, detail="Job not found")
+    return db.get_logs(job_id, limit=limit)
+
+
+@app.get("/api/jobs/{job_id}/logs/csv")
+async def download_log_csv(job_id: str):
+    """Download artifact log as CSV."""
+    if not db.get_job(job_id):
+        raise HTTPException(status_code=404, detail="Job not found")
+    csv_content = db.get_log_csv(job_id)
+    return StreamingResponse(
+        iter([csv_content]),
+        media_type="text/csv",
+        headers={"Content-Disposition": f'attachment; filename="job_{job_id[:8]}_log.csv"'},
+    )
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Server-Sent Events (SSE) for real-time updates
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.get("/api/jobs/{job_id}/events")
+async def job_events(job_id: str, request: Request):
+    """
+    SSE endpoint. Client subscribes and receives real-time events
+    as the worker processes artifacts.
+    """
+    if not db.get_job(job_id):
+        raise HTTPException(status_code=404, detail="Job not found")
+
+    queue = worker.emitter.subscribe(job_id)
+
+    async def event_generator():
+        try:
+            # Send initial state snapshot
+            job = db.get_job(job_id)
+            stats = db.get_job_stats(job_id)
+            if job:
+                job.update(stats)
+                yield _sse("job_status", {"job_id": job_id, "status": job["status"], "stats": stats})
+
+            while True:
+                if await request.is_disconnected():
+                    break
+
+                # Poll the queue with a short timeout
+                event = await asyncio.get_event_loop().run_in_executor(
+                    None, lambda: queue.get(timeout=1.0)
+                )
+
+                if event:
+                    yield _sse(event["type"], event["data"])
+                else:
+                    # Send a keepalive comment
+                    yield ": keepalive\n\n"
+
+                # Check if job is complete
+                job = db.get_job(job_id)
+                if job and job["status"] in ("done", "done_with_errors", "stopped", "failed"):
+                    # Give a moment for final events to flush
+                    await asyncio.sleep(0.2)
+                    # Drain remaining queue events
+                    while True:
+                        event = await asyncio.get_event_loop().run_in_executor(
+                            None, lambda: queue.get(timeout=0.1)
+                        )
+                        if event:
+                            yield _sse(event["type"], event["data"])
+                        else:
+                            break
+                    yield _sse("stream_end", {"job_id": job_id})
+                    break
+
+        except asyncio.CancelledError:
+            pass
+        finally:
+            worker.emitter.unsubscribe(job_id, queue)
+
+    return StreamingResponse(
+        event_generator(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "X-Accel-Buffering": "no",
+            "Connection": "keep-alive",
+        },
+    )
+
+
+def _sse(event_type: str, data: dict) -> str:
+    """Format a Server-Sent Event string."""
+    return f"event: {event_type}\ndata: {json.dumps(data)}\n\n"
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# File System Browsing Helper
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.get("/api/browse")
+async def browse_folder(path: str = ""):
+    """
+    Return directory listing for folder picking in the GUI.
+    Defaults to user home directory.
+    """
+    if not path:
+        path = str(Path.home())
+
+    target = Path(path)
+    if not target.exists() or not target.is_dir():
+        # Fallback to home
+        target = Path.home()
+
+    try:
+        entries = []
+        for entry in sorted(target.iterdir()):
+            try:
+                entries.append({
+                    "name": entry.name,
+                    "path": str(entry.absolute()),
+                    "is_dir": entry.is_dir(),
+                })
+            except PermissionError:
+                pass
+
+        # Add parent directory option
+        parent = str(target.parent.absolute())
+        return {
+            "current": str(target.absolute()),
+            "parent": parent if parent != str(target.absolute()) else None,
+            "entries": entries,
+        }
+    except PermissionError:
+        raise HTTPException(status_code=403, detail="Permission denied")
+
+
+@app.post("/api/mkdir")
+async def make_dir(req: BrowseFolderRequest):
+    """Create a new directory."""
+    if not req.path:
+        raise HTTPException(status_code=400, detail="path is required")
+    try:
+        Path(req.path).mkdir(parents=True, exist_ok=True)
+        return {"created": req.path}
+    except Exception as e:
+        raise HTTPException(status_code=400, detail=str(e))
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Config / Settings API
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.get("/api/config")
+async def get_config():
+    """Return non-sensitive defaults for the GUI."""
+    return {
+        "available_resolutions": config.AVAILABLE_RESOLUTIONS,
+        "artifact_types": config.ARTIFACT_TYPES,
+        "default_resolution": config.DEFAULT_VIDEO_RESOLUTION,
+        "delay_min": config.DELAY_MIN_SECONDS,
+        "delay_max": config.DELAY_MAX_SECONDS,
+        "max_retries": config.MAX_RETRIES,
+        "default_model": config.OPENROUTER_MODEL,
+        "summary_prompt": config.SUMMARY_SYSTEM_PROMPT,
+        "openrouter_base_url": config.OPENROUTER_BASE_URL,
+    }
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Health Check
+# ─────────────────────────────────────────────────────────────────────────────
+
+@app.get("/api/health")
+async def health():
+    return {"status": "ok", "version": "1.0.0"}
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Entry Point
+# ─────────────────────────────────────────────────────────────────────────────
+
+if __name__ == "__main__":
+    print(f"\n{'='*60}")
+    print("  YouTube SIPOC Tool")
+    print(f"  Web GUI: http://{config.API_HOST}:{config.API_PORT}")
+    print(f"  API:     http://{config.API_HOST}:{config.API_PORT}/api")
+    print(f"{'='*60}\n")
+    uvicorn.run(
+        "api_server:app",
+        host=config.API_HOST,
+        port=config.API_PORT,
+        reload=False,
+        log_level="info",
+    )
diff --git a/backend/batchtube-server.spec b/backend/batchtube-server.spec
new file mode 100644
index 0000000..acc0d23
--- /dev/null
+++ b/backend/batchtube-server.spec
@@ -0,0 +1,64 @@
+# batchtube-server.spec
+# PyInstaller spec to bundle the BatchTube FastAPI backend
+# Run: pyinstaller batchtube-server.spec
+
+import sys
+from pathlib import Path
+
+block_cipher = None
+
+a = Analysis(
+    ['api_server.py'],
+    pathex=[str(Path('.'). resolve())],
+    binaries=[],
+    datas=[
+        ('web', 'web'),
+        ('youtube_sipoc.db', '.'),
+    ],
+    hiddenimports=[
+        'uvicorn.logging',
+        'uvicorn.loops',
+        'uvicorn.loops.auto',
+        'uvicorn.protocols',
+        'uvicorn.protocols.http',
+        'uvicorn.protocols.http.auto',
+        'uvicorn.protocols.websockets',
+        'uvicorn.protocols.websockets.auto',
+        'uvicorn.lifespan',
+        'uvicorn.lifespan.on',
+        'fastapi',
+        'yt_dlp',
+    ],
+    hookspath=[],
+    hooksconfig={},
+    runtime_hooks=[],
+    excludes=[],
+    win_no_prefer_redirects=False,
+    win_private_assemblies=False,
+    cipher=block_cipher,
+    noarchive=False,
+)
+
+pyz = PYZ(a.pure, a.zipped_data, cipher=block_cipher)
+
+exe = EXE(
+    pyz,
+    a.scripts,
+    a.binaries,
+    a.zipfiles,
+    a.datas,
+    [],
+    name='batchtube-server',
+    debug=False,
+    bootloader_ignore_signals=False,
+    strip=False,
+    upx=True,
+    upx_exclude=[],
+    runtime_tmpdir=None,
+    console=False,  # No console window
+    disable_windowed_traceback=False,
+    argv_emulation=False,
+    target_arch=None,
+    codesign_identity=None,
+    entitlements_file=None,
+)
diff --git a/backend/config.py b/backend/config.py
new file mode 100644
index 0000000..0e2eca3
--- /dev/null
+++ b/backend/config.py
@@ -0,0 +1,38 @@
+"""
+config.py - Default configuration for YouTube SIPOC Tool
+"""
+import os
+
+# ── API Settings ──────────────────────────────────────────────────────────────
+OPENROUTER_API_KEY = os.environ.get("OPENROUTER_API_KEY", "")
+OPENROUTER_BASE_URL = "https://openrouter.ai/api/v1"
+OPENROUTER_MODEL = os.environ.get("OPENROUTER_MODEL", "openai/gpt-4o-mini")
+
+# ── Server Settings ───────────────────────────────────────────────────────────
+API_HOST = "127.0.0.1"
+API_PORT = 8765
+
+# ── Download / Bot Avoidance ──────────────────────────────────────────────────
+DELAY_MIN_SECONDS = 2.0   # minimum random pause between artifacts
+DELAY_MAX_SECONDS = 7.0   # maximum random pause between artifacts
+MAX_RETRIES = 3            # max retry attempts per failed artifact
+
+# ── Video Download ────────────────────────────────────────────────────────────
+DEFAULT_VIDEO_RESOLUTION = "1080"   # used if not specified by user
+AVAILABLE_RESOLUTIONS = ["best", "4320", "2160", "1440", "1080", "720", "480", "360", "240", "144"]
+
+# ── Summary ───────────────────────────────────────────────────────────────────
+SUMMARY_SYSTEM_PROMPT = (
+    "You are a helpful assistant that creates concise, well-structured summaries of video transcripts. "
+    "Include key points, main topics, and any actionable insights. Format with bullet points where appropriate."
+)
+SUMMARY_MAX_TOKENS = 1024
+
+# ── File Naming ───────────────────────────────────────────────────────────────
+MAX_TITLE_LENGTH = 80   # characters - truncate long titles in filenames
+
+# ── Database ──────────────────────────────────────────────────────────────────
+DB_PATH = os.path.join(os.path.dirname(__file__), "youtube_sipoc.db")
+
+# ── Artifact Types ────────────────────────────────────────────────────────────
+ARTIFACT_TYPES = ["title", "stats", "transcript", "summary", "thumbnail", "mp3", "mp4"]
diff --git a/backend/database.py b/backend/database.py
new file mode 100644
index 0000000..fcb63ed
--- /dev/null
+++ b/backend/database.py
@@ -0,0 +1,371 @@
+"""
+database.py - SQLite state management for YouTube SIPOC Tool
+Handles jobs, videos, artifacts, and logs with resume capability.
+"""
+import sqlite3
+import json
+import uuid
+from datetime import datetime
+from contextlib import contextmanager
+from typing import Optional, List, Dict, Any
+
+import config
+
+
+@contextmanager
+def get_conn():
+    """Context manager providing a thread-safe SQLite connection."""
+    conn = sqlite3.connect(config.DB_PATH, check_same_thread=False)
+    conn.row_factory = sqlite3.Row
+    conn.execute("PRAGMA journal_mode=WAL")  # WAL mode for concurrent reads
+    conn.execute("PRAGMA foreign_keys=ON")
+    try:
+        yield conn
+        conn.commit()
+    except Exception:
+        conn.rollback()
+        raise
+    finally:
+        conn.close()
+
+
+def init_db():
+    """Create all tables if they don't exist."""
+    with get_conn() as conn:
+        conn.executescript("""
+            CREATE TABLE IF NOT EXISTS jobs (
+                job_id      TEXT PRIMARY KEY,
+                status      TEXT NOT NULL DEFAULT 'pending',
+                created_at  TEXT NOT NULL,
+                updated_at  TEXT NOT NULL,
+                output_folder TEXT NOT NULL,
+                selections  TEXT NOT NULL,
+                resolution  TEXT NOT NULL DEFAULT 'best',
+                delay_min   REAL NOT NULL DEFAULT 2.0,
+                delay_max   REAL NOT NULL DEFAULT 7.0,
+                input_type  TEXT NOT NULL DEFAULT 'url',
+                input_value TEXT NOT NULL DEFAULT '',
+                summary_prompt TEXT NOT NULL DEFAULT '',
+                openrouter_model TEXT NOT NULL DEFAULT '',
+                total_videos INTEGER NOT NULL DEFAULT 0,
+                done_videos  INTEGER NOT NULL DEFAULT 0
+            );
+
+            CREATE TABLE IF NOT EXISTS videos (
+                id          INTEGER PRIMARY KEY AUTOINCREMENT,
+                job_id      TEXT NOT NULL REFERENCES jobs(job_id),
+                video_id    TEXT NOT NULL,
+                title       TEXT NOT NULL DEFAULT '',
+                url         TEXT NOT NULL,
+                position    INTEGER NOT NULL DEFAULT 0,
+                status      TEXT NOT NULL DEFAULT 'pending',
+                UNIQUE(job_id, video_id)
+            );
+
+            CREATE TABLE IF NOT EXISTS artifacts (
+                id              INTEGER PRIMARY KEY AUTOINCREMENT,
+                job_id          TEXT NOT NULL REFERENCES jobs(job_id),
+                video_id        TEXT NOT NULL,
+                artifact_type   TEXT NOT NULL,
+                status          TEXT NOT NULL DEFAULT 'pending',
+                file_path       TEXT,
+                error_msg       TEXT,
+                attempts        INTEGER NOT NULL DEFAULT 0,
+                started_at      TEXT,
+                finished_at     TEXT,
+                UNIQUE(job_id, video_id, artifact_type)
+            );
+
+            CREATE TABLE IF NOT EXISTS logs (
+                id          INTEGER PRIMARY KEY AUTOINCREMENT,
+                job_id      TEXT NOT NULL REFERENCES jobs(job_id),
+                video_id    TEXT,
+                artifact_type TEXT,
+                level       TEXT NOT NULL DEFAULT 'INFO',
+                message     TEXT NOT NULL,
+                timestamp   TEXT NOT NULL
+            );
+
+            CREATE INDEX IF NOT EXISTS idx_artifacts_job ON artifacts(job_id);
+            CREATE INDEX IF NOT EXISTS idx_videos_job ON videos(job_id);
+            CREATE INDEX IF NOT EXISTS idx_logs_job ON logs(job_id);
+        """)
+
+
+# ── Job Operations ────────────────────────────────────────────────────────────
+
+def create_job(
+    output_folder: str,
+    selections: List[str],
+    resolution: str,
+    delay_min: float,
+    delay_max: float,
+    input_type: str,
+    input_value: str,
+    summary_prompt: str,
+    openrouter_model: str,
+) -> str:
+    """Create a new job and return its job_id."""
+    job_id = str(uuid.uuid4())
+    now = datetime.utcnow().isoformat()
+    with get_conn() as conn:
+        conn.execute(
+            """INSERT INTO jobs
+               (job_id, status, created_at, updated_at, output_folder, selections,
+                resolution, delay_min, delay_max, input_type, input_value,
+                summary_prompt, openrouter_model)
+               VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)""",
+            (
+                job_id, "pending", now, now, output_folder,
+                json.dumps(selections), resolution,
+                delay_min, delay_max, input_type, input_value,
+                summary_prompt, openrouter_model,
+            ),
+        )
+    return job_id
+
+
+def get_job(job_id: str) -> Optional[Dict]:
+    with get_conn() as conn:
+        row = conn.execute("SELECT * FROM jobs WHERE job_id=?", (job_id,)).fetchone()
+        if row:
+            d = dict(row)
+            d["selections"] = json.loads(d["selections"])
+            return d
+    return None
+
+
+def list_jobs() -> List[Dict]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            "SELECT * FROM jobs ORDER BY created_at DESC"
+        ).fetchall()
+        result = []
+        for row in rows:
+            d = dict(row)
+            d["selections"] = json.loads(d["selections"])
+            result.append(d)
+        return result
+
+
+def update_job_status(job_id: str, status: str):
+    now = datetime.utcnow().isoformat()
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE jobs SET status=?, updated_at=? WHERE job_id=?",
+            (status, now, job_id),
+        )
+
+
+def update_job_progress(job_id: str, total: int, done: int):
+    now = datetime.utcnow().isoformat()
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE jobs SET total_videos=?, done_videos=?, updated_at=? WHERE job_id=?",
+            (total, done, now, job_id),
+        )
+
+
+def set_job_field(job_id: str, field: str, value: Any):
+    """Update a single field on the jobs table (safe whitelist)."""
+    allowed = {"status", "total_videos", "done_videos", "updated_at"}
+    if field not in allowed:
+        raise ValueError(f"Field '{field}' not allowed")
+    now = datetime.utcnow().isoformat()
+    with get_conn() as conn:
+        conn.execute(f"UPDATE jobs SET {field}=?, updated_at=? WHERE job_id=?", (value, now, job_id))
+
+
+# ── Video Operations ──────────────────────────────────────────────────────────
+
+def upsert_video(job_id: str, video_id: str, url: str, title: str = "", position: int = 0):
+    with get_conn() as conn:
+        conn.execute(
+            """INSERT INTO videos (job_id, video_id, title, url, position, status)
+               VALUES (?,?,?,?,?,'pending')
+               ON CONFLICT(job_id, video_id) DO UPDATE SET
+                 title=excluded.title,
+                 position=excluded.position""",
+            (job_id, video_id, title, url, position),
+        )
+
+
+def update_video_title(job_id: str, video_id: str, title: str):
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE videos SET title=? WHERE job_id=? AND video_id=?",
+            (title, job_id, video_id),
+        )
+
+
+def update_video_status(job_id: str, video_id: str, status: str):
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE videos SET status=? WHERE job_id=? AND video_id=?",
+            (status, job_id, video_id),
+        )
+
+
+def get_videos(job_id: str) -> List[Dict]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            "SELECT * FROM videos WHERE job_id=? ORDER BY position ASC",
+            (job_id,),
+        ).fetchall()
+        return [dict(r) for r in rows]
+
+
+# ── Artifact Operations ───────────────────────────────────────────────────────
+
+def upsert_artifact(job_id: str, video_id: str, artifact_type: str):
+    """Create artifact record if not exists (skip if already done/skipped)."""
+    with get_conn() as conn:
+        conn.execute(
+            """INSERT INTO artifacts (job_id, video_id, artifact_type, status)
+               VALUES (?,?,?,'pending')
+               ON CONFLICT(job_id, video_id, artifact_type) DO NOTHING""",
+            (job_id, video_id, artifact_type),
+        )
+
+
+def get_artifact(job_id: str, video_id: str, artifact_type: str) -> Optional[Dict]:
+    with get_conn() as conn:
+        row = conn.execute(
+            "SELECT * FROM artifacts WHERE job_id=? AND video_id=? AND artifact_type=?",
+            (job_id, video_id, artifact_type),
+        ).fetchone()
+        return dict(row) if row else None
+
+
+def artifact_is_done(job_id: str, video_id: str, artifact_type: str) -> bool:
+    """Return True if artifact already succeeded or was skipped."""
+    art = get_artifact(job_id, video_id, artifact_type)
+    return art is not None and art["status"] in ("done", "skipped")
+
+
+def update_artifact_status(
+    job_id: str,
+    video_id: str,
+    artifact_type: str,
+    status: str,
+    file_path: Optional[str] = None,
+    error_msg: Optional[str] = None,
+):
+    now = datetime.utcnow().isoformat()
+    with get_conn() as conn:
+        conn.execute(
+            """UPDATE artifacts
+               SET status=?, file_path=?, error_msg=?, finished_at=?,
+                   attempts = attempts + CASE WHEN ? IN ('done','failed','skipped') THEN 1 ELSE 0 END
+               WHERE job_id=? AND video_id=? AND artifact_type=?""",
+            (status, file_path, error_msg, now, status, job_id, video_id, artifact_type),
+        )
+
+
+def mark_artifact_started(job_id: str, video_id: str, artifact_type: str):
+    now = datetime.utcnow().isoformat()
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE artifacts SET status='running', started_at=? WHERE job_id=? AND video_id=? AND artifact_type=?",
+            (now, job_id, video_id, artifact_type),
+        )
+
+
+def get_artifacts_for_video(job_id: str, video_id: str) -> List[Dict]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            "SELECT * FROM artifacts WHERE job_id=? AND video_id=? ORDER BY id ASC",
+            (job_id, video_id),
+        ).fetchall()
+        return [dict(r) for r in rows]
+
+
+def get_all_artifacts(job_id: str) -> List[Dict]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            """SELECT a.*, v.title as video_title
+               FROM artifacts a
+               JOIN videos v ON a.job_id=v.job_id AND a.video_id=v.video_id
+               WHERE a.job_id=?
+               ORDER BY v.position ASC, a.artifact_type ASC""",
+            (job_id,),
+        ).fetchall()
+        return [dict(r) for r in rows]
+
+
+# ── Logging ───────────────────────────────────────────────────────────────────
+
+def add_log(
+    job_id: str,
+    message: str,
+    level: str = "INFO",
+    video_id: Optional[str] = None,
+    artifact_type: Optional[str] = None,
+):
+    now = datetime.utcnow().isoformat()
+    with get_conn() as conn:
+        conn.execute(
+            """INSERT INTO logs (job_id, video_id, artifact_type, level, message, timestamp)
+               VALUES (?,?,?,?,?,?)""",
+            (job_id, video_id, artifact_type, level, message, now),
+        )
+
+
+def get_logs(job_id: str, limit: int = 500) -> List[Dict]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            "SELECT * FROM logs WHERE job_id=? ORDER BY id DESC LIMIT ?",
+            (job_id, limit),
+        ).fetchall()
+        return [dict(r) for r in reversed(rows)]
+
+
+def get_log_csv(job_id: str) -> str:
+    """Export artifacts as CSV string for download."""
+    import io, csv
+    artifacts = get_all_artifacts(job_id)
+    output = io.StringIO()
+    writer = csv.writer(output)
+    writer.writerow(["video_id", "video_title", "artifact_type", "status", "file_path", "error_msg", "attempts", "started_at", "finished_at"])
+    for a in artifacts:
+        writer.writerow([
+            a.get("video_id", ""),
+            a.get("video_title", ""),
+            a.get("artifact_type", ""),
+            a.get("status", ""),
+            a.get("file_path", ""),
+            a.get("error_msg", ""),
+            a.get("attempts", 0),
+            a.get("started_at", ""),
+            a.get("finished_at", ""),
+        ])
+    return output.getvalue()
+
+
+def get_job_stats(job_id: str) -> Dict:
+    """Return aggregate counts for a job."""
+    with get_conn() as conn:
+        total_videos = conn.execute(
+            "SELECT COUNT(*) FROM videos WHERE job_id=?", (job_id,)
+        ).fetchone()[0]
+        done_videos = conn.execute(
+            "SELECT COUNT(*) FROM videos WHERE job_id=? AND status='done'", (job_id,)
+        ).fetchone()[0]
+        artifact_counts = conn.execute(
+            """SELECT status, COUNT(*) as cnt FROM artifacts WHERE job_id=? GROUP BY status""",
+            (job_id,),
+        ).fetchall()
+        counts = {row["status"]: row["cnt"] for row in artifact_counts}
+    return {
+        "total_videos": total_videos,
+        "done_videos": done_videos,
+        "artifact_done": counts.get("done", 0),
+        "artifact_failed": counts.get("failed", 0),
+        "artifact_pending": counts.get("pending", 0),
+        "artifact_running": counts.get("running", 0),
+        "artifact_skipped": counts.get("skipped", 0),
+    }
+
+
+# Initialize on import
+init_db()
diff --git a/backend/extractors.py b/backend/extractors.py
new file mode 100644
index 0000000..76e09fe
--- /dev/null
+++ b/backend/extractors.py
@@ -0,0 +1,749 @@
+"""
+extractors.py - Per-artifact extraction logic for YouTube SIPOC Tool
+Uses yt-dlp, youtube-transcript-api, and OpenRouter LLM.
+"""
+import os
+import re
+import json
+import time
+import logging
+import requests
+from pathlib import Path
+from typing import Optional, Dict, Tuple, List
+
+import yt_dlp
+from youtube_transcript_api import YouTubeTranscriptApi
+try:
+    from youtube_transcript_api import TranscriptsDisabled, NoTranscriptFound
+except ImportError:
+    # v1.x moved/renamed exceptions
+    TranscriptsDisabled = Exception
+    NoTranscriptFound = Exception
+from openai import OpenAI
+
+import config
+
+logger = logging.getLogger("extractors")
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Utilities
+# ─────────────────────────────────────────────────────────────────────────────
+
+def sanitize_filename(name: str, max_length: int = config.MAX_TITLE_LENGTH) -> str:
+    """Make a string safe to use in a file name."""
+    name = re.sub(r'[\\/:*?"<>|]', "_", name)
+    name = re.sub(r'\s+', "_", name.strip())
+    name = re.sub(r'_+', "_", name)
+    return name[:max_length].rstrip("._")
+
+
+def make_base_name(title: str, video_id: str) -> str:
+    """Return the standard base filename (no extension)."""
+    safe_title = sanitize_filename(title) if title else "untitled"
+    return f"{safe_title}__{video_id}"
+
+
+def ensure_dir(path: str):
+    Path(path).mkdir(parents=True, exist_ok=True)
+
+
+def video_url(video_id: str) -> str:
+    return f"https://www.youtube.com/watch?v={video_id}"
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Input Resolution: URL → list of video IDs
+# ─────────────────────────────────────────────────────────────────────────────
+
+def resolve_input(input_type: str, input_value: str) -> List[Dict]:
+    """
+    Resolve any input type to a list of dicts: {video_id, url, title, position}
+    Supported input_types: url, playlist, csv, handle
+    """
+    if input_type == "url":
+        return _resolve_single_url(input_value)
+    elif input_type == "playlist":
+        return _resolve_playlist(input_value)
+    elif input_type == "csv":
+        return _resolve_csv(input_value)
+    elif input_type == "handle":
+        return _resolve_handle(input_value)
+    else:
+        raise ValueError(f"Unknown input_type: {input_type}")
+
+
+def _extract_video_id(url: str) -> Optional[str]:
+    """Extract YouTube video ID from various URL formats."""
+    patterns = [
+        r'(?:v=|youtu\.be/|embed/|shorts/)([A-Za-z0-9_-]{11})',
+    ]
+    for pat in patterns:
+        m = re.search(pat, url)
+        if m:
+            return m.group(1)
+    return None
+
+
+def _resolve_single_url(url: str) -> List[Dict]:
+    """Handle single YouTube URL (video or short)."""
+    url = url.strip()
+    vid = _extract_video_id(url)
+    if vid:
+        return [{"video_id": vid, "url": video_url(vid), "title": "", "position": 0}]
+    # Fall through to yt-dlp for unusual URLs
+    return _resolve_with_ytdlp(url)
+
+
+def _resolve_playlist(playlist_url: str) -> List[Dict]:
+    return _resolve_with_ytdlp(playlist_url.strip())
+
+
+def _resolve_handle(handle: str) -> List[Dict]:
+    """Resolve @handle or channel URL to video list."""
+    handle = handle.strip()
+    if not handle.startswith("http"):
+        if handle.startswith("@"):
+            url = f"https://www.youtube.com/{handle}/videos"
+        else:
+            url = f"https://www.youtube.com/@{handle}/videos"
+    else:
+        url = handle
+    return _resolve_with_ytdlp(url)
+
+
+def _resolve_csv(csv_content: str) -> List[Dict]:
+    """Parse CSV content: each line may be a URL or video_id."""
+    lines = []
+    for line in csv_content.splitlines():
+        for part in line.split(","):
+            part = part.strip().strip('"').strip("'")
+            if part:
+                lines.append(part)
+    results = []
+    for i, entry in enumerate(lines):
+        if entry.startswith("http"):
+            items = _resolve_single_url(entry)
+            for item in items:
+                item["position"] = len(results)
+                results.append(item)
+        else:
+            # Treat as raw video_id
+            vid = entry[:11]
+            results.append({"video_id": vid, "url": video_url(vid), "title": "", "position": len(results)})
+    return results
+
+
+def _resolve_with_ytdlp(url: str) -> List[Dict]:
+    """Use yt-dlp to extract video list from any URL (playlist, channel, etc.)."""
+    captured_errors = []
+
+    class ErrorCollector:
+        """Custom yt-dlp logger to capture warning/error messages."""
+        def debug(self, msg): pass
+        def info(self, msg): pass
+        def warning(self, msg):
+            captured_errors.append(f"WARNING: {msg}")
+        def error(self, msg):
+            captured_errors.append(f"ERROR: {msg}")
+
+    ydl_opts = {
+        "quiet": True,
+        "extract_flat": True,
+        "skip_download": True,
+        "ignoreerrors": False,   # Don't swallow errors silently
+        "logger": ErrorCollector(),
+    }
+    results = []
+    info = None
+    try:
+        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+            info = ydl.extract_info(url, download=False)
+    except Exception as e:
+        captured_errors.append(str(e))
+
+    if not info:
+        # Build a meaningful error message
+        err_summary = "; ".join(captured_errors) if captured_errors else "No information returned by yt-dlp."
+        # Surface private/login-required as a clear user-facing error
+        lower = err_summary.lower()
+        if "private" in lower:
+            raise ValueError(
+                f"This playlist or channel is PRIVATE. "
+                f"YouTube requires you to be logged in to access it. "
+                f"To process private content, add your YouTube cookies file path in Settings. "
+                f"Details: {err_summary}"
+            )
+        elif "sign in" in lower or "login" in lower or "confirm your age" in lower:
+            raise ValueError(
+                f"YouTube requires authentication (sign-in or age verification). "
+                f"Add your YouTube cookies file path in Settings. "
+                f"Details: {err_summary}"
+            )
+        elif "not exist" in lower or "removed" in lower or "unavailable" in lower:
+            raise ValueError(f"Content not available or removed. Details: {err_summary}")
+        else:
+            raise ValueError(f"Could not resolve URL '{url}'. Details: {err_summary}")
+
+    entries = info.get("entries") or [info]
+    for i, entry in enumerate(entries):
+        if not entry:
+            continue
+        vid = entry.get("id") or entry.get("video_id")
+        if not vid:
+            continue
+        title = entry.get("title", "")
+        results.append({
+            "video_id": vid,
+            "url": video_url(vid),
+            "title": title,
+            "position": i,
+        })
+
+    if not results:
+        raise ValueError(
+            f"URL resolved successfully but contained 0 videos. "
+            f"The playlist/channel may be empty. URL: {url}"
+        )
+
+    return results
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Metadata Fetch
+# ─────────────────────────────────────────────────────────────────────────────
+
+def fetch_metadata(video_id: str) -> Dict:
+    """Fetch full video metadata using yt-dlp."""
+    ydl_opts = {
+        "quiet": True,
+        "skip_download": True,
+        "ignoreerrors": True,
+    }
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+        info = ydl.extract_info(video_url(video_id), download=False)
+    return info or {}
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact: Title
+# ─────────────────────────────────────────────────────────────────────────────
+
+def extract_title(video_id: str, meta: Dict, output_folder: str) -> Tuple[str, str]:
+    """
+    Save title as a .txt file.
+    Returns (file_path, title_string)
+    """
+    title = meta.get("title", "")
+    base = make_base_name(title, video_id)
+    file_path = os.path.join(output_folder, f"{base}__title.txt")
+    ensure_dir(output_folder)
+    with open(file_path, "w", encoding="utf-8") as f:
+        f.write(f"Title: {title}\n")
+        f.write(f"Video ID: {video_id}\n")
+        f.write(f"URL: {video_url(video_id)}\n")
+    return file_path, title
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact: Stats
+# ─────────────────────────────────────────────────────────────────────────────
+
+def extract_stats(video_id: str, meta: Dict, output_folder: str) -> str:
+    """Save video stats + full metadata as a human-readable .txt file."""
+    title = meta.get("title", "")
+    base = make_base_name(title, video_id)
+    file_path = os.path.join(output_folder, f"{base}__stats.txt")
+    ensure_dir(output_folder)
+
+    # Format helpers
+    def fmt_num(n) -> str:
+        if n is None:
+            return "N/A"
+        try:
+            return f"{int(n):,}"
+        except Exception:
+            return str(n)
+
+    def fmt_duration(secs) -> str:
+        if not secs:
+            return "N/A"
+        try:
+            secs = int(secs)
+            h, rem = divmod(secs, 3600)
+            m, s = divmod(rem, 60)
+            return f"{h}h {m}m {s}s" if h else f"{m}m {s}s"
+        except Exception:
+            return str(secs)
+
+    def fmt_date(d) -> str:
+        if not d:
+            return "N/A"
+        try:
+            d = str(d)
+            return f"{d[:4]}-{d[4:6]}-{d[6:8]}"
+        except Exception:
+            return str(d)
+
+    tags = meta.get("tags") or []
+    categories = meta.get("categories") or []
+    chapters = meta.get("chapters") or []
+    description = meta.get("description", "") or ""
+    formats = meta.get("formats") or []
+    subtitles = meta.get("subtitles") or {}
+    automatic_captions = meta.get("automatic_captions") or {}
+
+    with open(file_path, "w", encoding="utf-8") as f:
+        f.write("=" * 70 + "\n")
+        f.write(f"  VIDEO STATS & METADATA\n")
+        f.write("=" * 70 + "\n\n")
+
+        f.write(f"Title            : {meta.get('title', '')}\n")
+        f.write(f"Video ID         : {video_id}\n")
+        f.write(f"URL              : {video_url(video_id)}\n")
+        f.write(f"Webpage URL      : {meta.get('webpage_url', '')}\n\n")
+
+        f.write("── Channel ──────────────────────────────────────────────────\n")
+        f.write(f"Channel          : {meta.get('channel', meta.get('uploader', ''))}\n")
+        f.write(f"Channel ID       : {meta.get('channel_id', '')}\n")
+        f.write(f"Channel URL      : {meta.get('channel_url', meta.get('uploader_url', ''))}\n")
+        f.write(f"Subscribers      : {fmt_num(meta.get('channel_follower_count'))}\n\n")
+
+        f.write("── Content ──────────────────────────────────────────────────\n")
+        f.write(f"Upload Date      : {fmt_date(meta.get('upload_date'))}\n")
+        f.write(f"Duration         : {fmt_duration(meta.get('duration'))} ({meta.get('duration_string', '')})\n")
+        f.write(f"Age Limit        : {meta.get('age_limit', 0)}\n")
+        f.write(f"Is Live          : {meta.get('is_live', False)}\n")
+        f.write(f"Was Live         : {meta.get('was_live', False)}\n")
+        f.write(f"Live Status      : {meta.get('live_status', '')}\n\n")
+
+        f.write("── Engagement ───────────────────────────────────────────────\n")
+        f.write(f"Views            : {fmt_num(meta.get('view_count'))}\n")
+        f.write(f"Likes            : {fmt_num(meta.get('like_count'))}\n")
+        f.write(f"Comments         : {fmt_num(meta.get('comment_count'))}\n")
+        f.write(f"Average Rating   : {meta.get('average_rating', 'N/A')}\n\n")
+
+        f.write("── Media Info ───────────────────────────────────────────────\n")
+        f.write(f"Thumbnail URL    : {meta.get('thumbnail', '')}\n")
+        # Best format info
+        best_fmt = next((fmt for fmt in reversed(formats) if fmt.get("vcodec") not in (None, "none")), None)
+        if best_fmt:
+            f.write(f"Best Resolution  : {best_fmt.get('width', '?')}x{best_fmt.get('height', '?')} "
+                    f"@ {best_fmt.get('fps', '?')}fps\n")
+            f.write(f"Best Video Codec : {best_fmt.get('vcodec', 'N/A')}\n")
+            f.write(f"Best Audio Codec : {best_fmt.get('acodec', 'N/A')}\n")
+        f.write(f"Available Formats: {len(formats)}\n\n")
+
+        f.write("── Classification ───────────────────────────────────────────\n")
+        f.write(f"Categories       : {', '.join(categories) if categories else 'N/A'}\n")
+        if tags:
+            tag_str = ", ".join(tags[:30])
+            if len(tags) > 30:
+                tag_str += f" ... (+{len(tags)-30} more)"
+            f.write(f"Tags             : {tag_str}\n")
+        else:
+            f.write(f"Tags             : N/A\n")
+        f.write("\n")
+
+        # Subtitles / transcript languages
+        all_sub_langs = sorted(set(list(subtitles.keys()) + list(automatic_captions.keys())))
+        if all_sub_langs:
+            f.write("── Subtitles / Captions ─────────────────────────────────────\n")
+            f.write(f"Languages ({len(all_sub_langs)}): {', '.join(all_sub_langs[:20])}")
+            if len(all_sub_langs) > 20:
+                f.write(f" ... (+{len(all_sub_langs)-20} more)")
+            f.write("\n\n")
+
+        # Chapters
+        if chapters:
+            f.write("── Chapters ─────────────────────────────────────────────────\n")
+            for ch in chapters:
+                start = ch.get("start_time", 0)
+                m, s = divmod(int(start), 60)
+                h, m = divmod(m, 60)
+                ts = f"{h:02d}:{m:02d}:{s:02d}" if h else f"{m:02d}:{s:02d}"
+                f.write(f"  [{ts}] {ch.get('title', '')}\n")
+            f.write("\n")
+
+        # Description
+        if description:
+            f.write("── Description ──────────────────────────────────────────────\n")
+            # Limit description to 3000 chars to keep the file manageable
+            if len(description) > 3000:
+                f.write(description[:3000] + "\n... [truncated]\n")
+            else:
+                f.write(description + "\n")
+            f.write("\n")
+
+        f.write("── Extraction Info ──────────────────────────────────────────\n")
+        f.write(f"Extracted At     : {time.strftime('%Y-%m-%d %H:%M:%S UTC', time.gmtime())}\n")
+        f.write(f"Extractor        : {meta.get('extractor', 'youtube')}\n")
+        f.write(f"Extractor Key    : {meta.get('extractor_key', 'Youtube')}\n")
+        f.write("=" * 70 + "\n")
+
+    return file_path
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact: Transcript
+# ─────────────────────────────────────────────────────────────────────────────
+
+def _get_snippet_text(snippet) -> str:
+    """Safely extract text from a transcript snippet (dict or object)."""
+    if isinstance(snippet, dict):
+        return snippet.get("text", "")
+    return getattr(snippet, "text", "") or ""
+
+
+def _get_snippet_start(snippet) -> float:
+    """Safely extract start time from a transcript snippet (dict or object)."""
+    if isinstance(snippet, dict):
+        return snippet.get("start", 0.0)
+    return getattr(snippet, "start", 0.0) or 0.0
+
+
+def _fetch_transcript(video_id: str):
+    """
+    Fetch transcript snippets using youtube-transcript-api v1.x.
+    v1.x API: YouTubeTranscriptApi().list(video_id) and .fetch(video_id, languages=[...])
+    Returns (FetchedTranscript iterable, errors_list) — iterable may be None on failure.
+    """
+    errors = []
+    api = YouTubeTranscriptApi()
+
+    # Fast path: try direct fetch with preferred languages
+    for lang_group in [["en", "en-US", "en-GB"], ["en-US"], ["en-GB"]]:
+        try:
+            fetched = api.fetch(video_id, languages=lang_group)
+            return fetched, errors
+        except Exception:
+            pass
+
+    # Slower path: list all available transcripts and pick the best
+    try:
+        transcript_list = api.list(video_id)
+    except Exception as e:
+        errors.append(f"list error: {e}")
+        # Last-ditch: fetch with no language preference
+        try:
+            fetched = api.fetch(video_id)
+            return fetched, errors
+        except Exception as e2:
+            errors.append(f"fetch error: {e2}")
+            return None, errors
+
+    # Prefer any manual transcript
+    all_transcripts = list(transcript_list)
+    manual = [t for t in all_transcripts if not getattr(t, "is_generated", True)]
+    auto   = [t for t in all_transcripts if getattr(t, "is_generated", False)]
+
+    for t in (manual + auto):
+        try:
+            # Try fetching via the transcript object's own fetch method
+            fetched = t.fetch()
+            return fetched, errors
+        except Exception as e:
+            errors.append(f"t.fetch() error for {getattr(t, 'language_code','?')}: {e}")
+            continue
+
+    errors.append("No fetchable transcripts found")
+    return None, errors
+
+
+def extract_transcript(video_id: str, meta: Dict, output_folder: str) -> str:
+    """
+    Download transcript using youtube-transcript-api.
+    Falls back to auto-generated captions. Saves as .txt with timestamps.
+    """
+    title = meta.get("title", "")
+    base = make_base_name(title, video_id)
+    file_path = os.path.join(output_folder, f"{base}__transcript.txt")
+    ensure_dir(output_folder)
+
+    transcript, errors = _fetch_transcript(video_id)
+
+    if not transcript:
+        raise RuntimeError(f"Could not retrieve transcript for {video_id}. Errors: {'; '.join(errors)}")
+
+    # Write formatted transcript
+    with open(file_path, "w", encoding="utf-8") as f:
+        f.write(f"Transcript: {meta.get('title', video_id)}\n")
+        f.write(f"Video ID: {video_id}\n")
+        f.write(f"URL: {video_url(video_id)}\n")
+        f.write("=" * 60 + "\n\n")
+        for entry in transcript:
+            start = _get_snippet_start(entry)
+            text = _get_snippet_text(entry).replace("\n", " ")
+            minutes, seconds = divmod(int(start), 60)
+            hours, minutes = divmod(minutes, 60)
+            ts = f"[{hours:02d}:{minutes:02d}:{seconds:02d}]" if hours else f"[{minutes:02d}:{seconds:02d}]"
+            f.write(f"{ts} {text}\n")
+
+    return file_path
+
+
+def get_transcript_text(video_id: str) -> str:
+    """Get plain transcript text for summarization."""
+    try:
+        transcript, _ = _fetch_transcript(video_id)
+        if transcript:
+            return " ".join(_get_snippet_text(e) for e in transcript)
+    except Exception:
+        pass
+    return ""
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact: Summary
+# ─────────────────────────────────────────────────────────────────────────────
+
+def extract_summary(
+    video_id: str,
+    meta: Dict,
+    output_folder: str,
+    api_key: str,
+    model: str,
+    system_prompt: str,
+) -> str:
+    """Generate a summary using OpenRouter LLM and save as .txt."""
+    if not api_key:
+        raise RuntimeError("OpenRouter API key is not configured. Set it in the Settings panel.")
+
+    title = meta.get("title", "")
+    base = make_base_name(title, video_id)
+    file_path = os.path.join(output_folder, f"{base}__summary.txt")
+    ensure_dir(output_folder)
+
+    # Get transcript text for summarization
+    transcript_text = get_transcript_text(video_id)
+    if not transcript_text:
+        transcript_text = meta.get("description", "")
+    if not transcript_text:
+        raise RuntimeError(f"No transcript or description available for summarization of {video_id}")
+
+    # Truncate very long transcripts to ~12000 chars (token budget)
+    max_chars = 12000
+    if len(transcript_text) > max_chars:
+        transcript_text = transcript_text[:max_chars] + "\n\n[Transcript truncated for length]"
+
+    client = OpenAI(
+        api_key=api_key,
+        base_url=config.OPENROUTER_BASE_URL,
+    )
+
+    prompt = (
+        f"Title: {title}\n"
+        f"Video ID: {video_id}\n"
+        f"URL: {video_url(video_id)}\n\n"
+        f"Transcript:\n{transcript_text}"
+    )
+
+    response = client.chat.completions.create(
+        model=model or config.OPENROUTER_MODEL,
+        messages=[
+            {"role": "system", "content": system_prompt or config.SUMMARY_SYSTEM_PROMPT},
+            {"role": "user", "content": prompt},
+        ],
+        max_tokens=config.SUMMARY_MAX_TOKENS,
+    )
+
+    summary_text = response.choices[0].message.content or ""
+
+    with open(file_path, "w", encoding="utf-8") as f:
+        f.write(f"Summary: {title}\n")
+        f.write(f"Video ID: {video_id}\n")
+        f.write(f"URL: {video_url(video_id)}\n")
+        f.write("=" * 60 + "\n\n")
+        f.write(summary_text)
+        f.write("\n")
+
+    return file_path
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact: Thumbnail
+# ─────────────────────────────────────────────────────────────────────────────
+
+def extract_thumbnail(video_id: str, meta: Dict, output_folder: str) -> str:
+    """Download the highest-resolution thumbnail available."""
+    title = meta.get("title", "")
+    base = make_base_name(title, video_id)
+    ensure_dir(output_folder)
+
+    # Build priority list of thumbnail URLs
+    thumbnails = meta.get("thumbnails", [])
+    # Sort by resolution (width * height), highest first
+    thumbnails_sorted = sorted(
+        [t for t in thumbnails if t.get("url")],
+        key=lambda t: (t.get("width", 0) or 0) * (t.get("height", 0) or 0),
+        reverse=True,
+    )
+
+    # Fallback URLs
+    fallback_urls = [
+        f"https://img.youtube.com/vi/{video_id}/maxresdefault.jpg",
+        f"https://img.youtube.com/vi/{video_id}/sddefault.jpg",
+        f"https://img.youtube.com/vi/{video_id}/hqdefault.jpg",
+        f"https://img.youtube.com/vi/{video_id}/mqdefault.jpg",
+        f"https://img.youtube.com/vi/{video_id}/default.jpg",
+    ]
+
+    candidate_urls = [t["url"] for t in thumbnails_sorted] + fallback_urls
+
+    for url in candidate_urls:
+        try:
+            response = requests.get(url, timeout=15)
+            if response.status_code == 200 and len(response.content) > 1000:
+                # Determine extension from content-type or URL
+                ct = response.headers.get("content-type", "")
+                if "webp" in ct or url.endswith(".webp"):
+                    ext = "webp"
+                elif "png" in ct or url.endswith(".png"):
+                    ext = "png"
+                else:
+                    ext = "jpg"
+                file_path = os.path.join(output_folder, f"{base}__thumbnail.{ext}")
+                with open(file_path, "wb") as f:
+                    f.write(response.content)
+                return file_path
+        except Exception:
+            continue
+
+    raise RuntimeError(f"Could not download thumbnail for {video_id}")
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact: MP3
+# ─────────────────────────────────────────────────────────────────────────────
+
+def extract_mp3(video_id: str, meta: Dict, output_folder: str) -> str:
+    """Download audio and convert to MP3 using yt-dlp + ffmpeg."""
+    title = meta.get("title", "")
+    base = make_base_name(title, video_id)
+    ensure_dir(output_folder)
+
+    ffmpeg_dir = _find_ffmpeg()
+
+    # yt-dlp will append extension itself; we use outtmpl without extension
+    out_template = os.path.join(output_folder, f"{base}__audio")
+    expected_path = os.path.join(output_folder, f"{base}__audio.mp3")
+
+    ydl_opts = {
+        "quiet": True,
+        "no_warnings": True,
+        "format": "bestaudio/best",
+        "outtmpl": out_template,
+        "postprocessors": [
+            {
+                "key": "FFmpegExtractAudio",
+                "preferredcodec": "mp3",
+                "preferredquality": "192",
+            }
+        ],
+        "ignoreerrors": False,
+    }
+
+    if ffmpeg_dir:
+        ydl_opts["ffmpeg_location"] = ffmpeg_dir
+
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+        ydl.download([video_url(video_id)])
+
+    # Rename to standard naming convention
+    if os.path.exists(expected_path):
+        final_path = os.path.join(output_folder, f"{base}__mp3.mp3")
+        if not os.path.exists(final_path):
+            os.rename(expected_path, final_path)
+        return final_path
+
+    # Search for any mp3 file created in this run
+    for fname in os.listdir(output_folder):
+        if fname.startswith(base) and fname.endswith(".mp3"):
+            final_path = os.path.join(output_folder, f"{base}__mp3.mp3")
+            src = os.path.join(output_folder, fname)
+            if src != final_path and not os.path.exists(final_path):
+                os.rename(src, final_path)
+            return final_path
+
+    raise RuntimeError(f"MP3 file not found after download for {video_id}")
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact: MP4
+# ─────────────────────────────────────────────────────────────────────────────
+
+def _find_ffmpeg() -> Optional[str]:
+    """Find ffmpeg binary location."""
+    # Common Windows locations
+    candidates = [
+        r"C:\ProgramData\chocolatey\bin",
+        r"C:\ffmpeg\bin",
+        r"C:\Program Files\ffmpeg\bin",
+        r"C:\Program Files (x86)\ffmpeg\bin",
+    ]
+    for folder in candidates:
+        if os.path.isfile(os.path.join(folder, "ffmpeg.exe")):
+            return folder
+    # Try PATH
+    import shutil
+    found = shutil.which("ffmpeg")
+    if found:
+        return str(Path(found).parent)
+    return None
+
+
+def extract_mp4(video_id: str, meta: Dict, output_folder: str, resolution: str = "best") -> str:
+    """Download video as MP4 using yt-dlp. Explicitly uses ffmpeg for merging."""
+    title = meta.get("title", "")
+    base = make_base_name(title, video_id)
+    ensure_dir(output_folder)
+
+    ffmpeg_dir = _find_ffmpeg()
+
+    # Build format selector — always require vcodec to avoid audio-only streams
+    if resolution == "best":
+        fmt = (
+            "bestvideo[vcodec^=avc1][ext=mp4]+bestaudio[ext=m4a]"
+            "/bestvideo[ext=mp4]+bestaudio[ext=m4a]"
+            "/bestvideo+bestaudio"
+            "/best[vcodec!=none]"
+            "/best"
+        )
+    else:
+        fmt = (
+            f"bestvideo[vcodec^=avc1][height<={resolution}][ext=mp4]+bestaudio[ext=m4a]"
+            f"/bestvideo[height<={resolution}][ext=mp4]+bestaudio[ext=m4a]"
+            f"/bestvideo[height<={resolution}]+bestaudio"
+            f"/best[height<={resolution}][vcodec!=none]"
+            f"/best[height<={resolution}]"
+            f"/best[vcodec!=none]"
+            f"/best"
+        )
+
+    out_template = os.path.join(output_folder, f"{base}__mp4.%(ext)s")
+
+    ydl_opts = {
+        "quiet": True,
+        "no_warnings": True,
+        "format": fmt,
+        "outtmpl": out_template,
+        "merge_output_format": "mp4",
+        "ignoreerrors": False,
+        "postprocessors": [
+            {"key": "FFmpegVideoConvertor", "preferedformat": "mp4"},
+        ],
+    }
+
+    # Explicitly tell yt-dlp where ffmpeg is
+    if ffmpeg_dir:
+        ydl_opts["ffmpeg_location"] = ffmpeg_dir
+
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+        ydl.download([video_url(video_id)])
+
+    # Find the downloaded file
+    for fname in sorted(os.listdir(output_folder)):
+        if fname.startswith(f"{base}__mp4") and fname.endswith(".mp4"):
+            return os.path.join(output_folder, fname)
+
+    raise RuntimeError(f"MP4 file not found after download for {video_id}. ffmpeg_dir={ffmpeg_dir}")
diff --git a/backend/requirements.txt b/backend/requirements.txt
new file mode 100644
index 0000000..5e5b489
--- /dev/null
+++ b/backend/requirements.txt
@@ -0,0 +1,30 @@
+
+# Core
+fastapi>=0.110.0
+uvicorn[standard]>=0.29.0
+sse-starlette>=1.8.2
+python-multipart>=0.0.9
+
+# YouTube downloading
+yt-dlp>=2024.3.10
+
+# Transcripts
+youtube-transcript-api>=0.6.2
+
+# HTTP / utilities
+requests>=2.31.0
+aiofiles>=23.2.1
+
+# LLM (OpenRouter via OpenAI-compatible API)
+openai>=1.14.0
+
+# Database
+# sqlite3 is built-in to Python
+
+# MCP
+mcp>=1.0.0
+
+# Misc
+python-slugify>=8.0.4
+pydantic>=2.6.0
+tqdm>=4.66.2
diff --git a/backend/web/app.js b/backend/web/app.js
new file mode 100644
index 0000000..2c487ba
Binary files /dev/null and b/backend/web/app.js differ
diff --git a/backend/web/index.html b/backend/web/index.html
new file mode 100644
index 0000000..64e343c
--- /dev/null
+++ b/backend/web/index.html
@@ -0,0 +1,265 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1.0" />
+  <title>YouTube SIPOC Tool</title>
+  <link rel="stylesheet" href="/style.css" />
+</head>
+<body>
+
+<!-- ═══════════════════════════ HEADER ═══════════════════════════ -->
+<header class="app-header">
+  <div class="app-logo">
+    <svg width="28" height="28" viewBox="0 0 24 24" fill="currentColor">
+      <path d="M19.59 6.69a4.83 4.83 0 01-3.77-4.25V2h-3.45v13.67a2.89 2.89 0 01-2.88 2.5 2.89 2.89 0 01-2.89-2.89 2.89 2.89 0 012.89-2.89c.28 0 .54.04.79.1V9.01a6.33 6.33 0 00-.79-.05 6.34 6.34 0 00-6.34 6.34 6.34 6.34 0 006.34 6.34 6.34 6.34 0 006.33-6.34v-7a8.15 8.15 0 004.77 1.52v-3.4a4.85 4.85 0 01-1-.09z"/>
+    </svg>
+    <span class="app-title">YouTube <span>SIPOC</span></span>
+  </div>
+  <nav class="header-nav">
+    <button class="nav-btn active" onclick="showView('jobs')" id="nav-jobs">Jobs</button>
+    <button class="nav-btn" onclick="showView('new')" id="nav-new">+ New Job</button>
+    <button class="nav-btn" onclick="showView('settings')" id="nav-settings">⚙ Settings</button>
+  </nav>
+</header>
+
+<div class="app-body">
+
+  <!-- ═══════════════════════ SIDEBAR ═══════════════════════ -->
+  <aside class="sidebar">
+    <div class="sidebar-section">
+      <button class="new-job-btn" onclick="showView('new')">
+        <svg width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2.5"><line x1="12" y1="5" x2="12" y2="19"/><line x1="5" y1="12" x2="19" y2="12"/></svg>
+        New Extraction Job
+      </button>
+    </div>
+    <div class="sidebar-section" style="flex:1; overflow-y:auto; border-bottom:none;">
+      <div class="sidebar-label">Recent Jobs</div>
+      <div class="job-list" id="sidebar-job-list">
+        <div class="text-muted" style="font-size:12px;padding:8px 4px;">No jobs yet</div>
+      </div>
+    </div>
+  </aside>
+
+  <!-- ═══════════════════════ MAIN CONTENT ═══════════════════════ -->
+  <main class="main-content">
+
+    <!-- ─── VIEW: NEW JOB ─── -->
+    <div id="view-new" class="hidden">
+      <div class="card">
+        <div class="card-header">
+          <svg width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="var(--accent)" stroke-width="2"><circle cx="12" cy="12" r="10"/><line x1="12" y1="8" x2="12" y2="16"/><line x1="8" y1="12" x2="16" y2="12"/></svg>
+          <div>
+            <div class="card-title">New Extraction Job</div>
+            <div class="card-subtitle">Configure YouTube source and what to extract</div>
+          </div>
+        </div>
+
+        <!-- Input Type -->
+        <div class="form-group">
+          <label class="form-label">Input Type</label>
+          <div class="type-pills">
+            <div class="type-pill selected" data-type="url" onclick="selectInputType('url')">🔗 Single URL</div>
+            <div class="type-pill" data-type="playlist" onclick="selectInputType('playlist')">📋 Playlist</div>
+            <div class="type-pill" data-type="handle" onclick="selectInputType('handle')">👤 Channel Handle</div>
+            <div class="type-pill" data-type="csv" onclick="selectInputType('csv')">📄 CSV List</div>
+          </div>
+        </div>
+
+        <!-- Input Value -->
+        <div class="form-group" id="input-url-group">
+          <label class="form-label" id="input-label">YouTube URL</label>
+          <input type="text" class="form-input" id="input-value"
+            placeholder="https://www.youtube.com/watch?v=..." />
+        </div>
+
+        <!-- CSV textarea (only shown for csv type) -->
+        <div class="form-group hidden" id="input-csv-group">
+          <label class="form-label">CSV / List of URLs (one per line or comma-separated)</label>
+          <textarea class="form-textarea" id="input-csv-value" rows="5"
+            placeholder="https://youtu.be/abc123&#10;https://youtu.be/xyz456&#10;dQw4w9WgXcQ"></textarea>
+        </div>
+
+        <hr class="divider" />
+
+        <!-- Artifact Selection -->
+        <div class="form-group">
+          <label class="form-label">What to Extract (select all that apply)</label>
+          <div class="artifact-grid" id="artifact-grid">
+            <!-- populated by JS -->
+          </div>
+        </div>
+
+        <hr class="divider" />
+
+        <!-- Output Folder -->
+        <div class="form-group">
+          <label class="form-label">Output Folder</label>
+          <div class="folder-display" id="folder-display" onclick="openFolderBrowser()">
+            <span>📁</span>
+            <span class="folder-path" id="folder-path-text">Click to choose a folder…</span>
+          </div>
+          <input type="hidden" id="output-folder" value="" />
+        </div>
+
+        <!-- Resolution (shown only when mp4 selected) -->
+        <div class="form-group hidden" id="resolution-group">
+          <label class="form-label">MP4 Resolution</label>
+          <select class="form-select" id="resolution">
+            <option value="best">Best Available</option>
+            <option value="2160">4K (2160p)</option>
+            <option value="1440">2K (1440p)</option>
+            <option value="1080" selected>1080p</option>
+            <option value="720">720p</option>
+            <option value="480">480p</option>
+            <option value="360">360p</option>
+            <option value="240">240p</option>
+            <option value="144">144p</option>
+          </select>
+        </div>
+
+        <!-- Bot Avoidance Delays -->
+        <div class="form-group">
+          <label class="form-label">Random Delay Between Downloads (seconds, for bot detection avoidance)</label>
+          <div class="range-row">
+            <div class="form-group" style="margin:0">
+              <label class="form-label" style="font-size:11px">Min Delay</label>
+              <input type="number" class="form-input" id="delay-min" value="2" min="0" max="60" step="0.5" style="width:100px" />
+            </div>
+            <div class="form-group" style="margin:0">
+              <label class="form-label" style="font-size:11px">Max Delay</label>
+              <input type="number" class="form-input" id="delay-max" value="7" min="0" max="120" step="0.5" style="width:100px" />
+            </div>
+          </div>
+        </div>
+
+        <hr class="divider" />
+
+        <!-- API Settings -->
+        <div class="form-group">
+          <label class="form-label">OpenRouter API Key <span class="text-muted">(required for summaries)</span></label>
+          <input type="password" class="form-input" id="api-key"
+            placeholder="sk-or-v1-..." autocomplete="off" />
+          <div style="font-size:11px;color:var(--text-muted);margin-top:4px;">
+            Not stored — only used for this session's summary requests
+          </div>
+        </div>
+
+        <div class="form-group">
+          <label class="form-label">OpenRouter Model</label>
+          <input type="text" class="form-input" id="or-model"
+            value="openai/gpt-4o-mini"
+            placeholder="openai/gpt-4o-mini" />
+        </div>
+
+        <div class="form-group">
+          <label class="form-label">Summary Prompt <span class="text-muted">(leave blank for default)</span></label>
+          <textarea class="form-textarea" id="summary-prompt" rows="3"
+            placeholder="Custom system prompt for LLM summarization (optional)…"></textarea>
+        </div>
+
+        <div class="btn-row">
+          <button class="btn btn-primary" onclick="createJob()" id="create-btn">
+            <svg width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2.5"><polyline points="5 12 12 5 19 12"/><polyline points="12 19 12 5"/></svg>
+            Start Extraction
+          </button>
+          <button class="btn btn-outline" onclick="showView('jobs')">Cancel</button>
+        </div>
+      </div>
+    </div>
+
+    <!-- ─── VIEW: JOB DETAIL ─── -->
+    <div id="view-jobs">
+      <div id="job-detail">
+        <!-- Welcome / empty state -->
+        <div class="empty-state">
+          <div class="empty-state-icon">🎬</div>
+          <div class="empty-state-title">YouTube SIPOC Tool</div>
+          <div class="empty-state-body">
+            Extract titles, stats, transcripts, summaries, thumbnails, MP3, and MP4 from YouTube videos.<br/><br/>
+            Click <strong>+ New Job</strong> to get started.
+          </div>
+        </div>
+      </div>
+    </div>
+
+    <!-- ─── VIEW: SETTINGS ─── -->
+    <div id="view-settings" class="hidden">
+      <div class="card">
+        <div class="card-header">
+          <div class="card-title">⚙ Settings & Defaults</div>
+          <div class="card-subtitle">These defaults are used when creating new jobs</div>
+        </div>
+        <div class="settings-grid">
+          <div class="form-group">
+            <label class="form-label">Default Delay Min (s)</label>
+            <input type="number" class="form-input" id="s-delay-min" value="2" min="0" />
+          </div>
+          <div class="form-group">
+            <label class="form-label">Default Delay Max (s)</label>
+            <input type="number" class="form-input" id="s-delay-max" value="7" min="0" />
+          </div>
+          <div class="form-group">
+            <label class="form-label">Default MP4 Resolution</label>
+            <select class="form-select" id="s-resolution">
+              <option value="best">Best</option>
+              <option value="1080" selected>1080p</option>
+              <option value="720">720p</option>
+              <option value="480">480p</option>
+            </select>
+          </div>
+          <div class="form-group">
+            <label class="form-label">Default OpenRouter Model</label>
+            <input type="text" class="form-input" id="s-model" value="openai/gpt-4o-mini" />
+          </div>
+        </div>
+        <div class="form-group">
+          <label class="form-label">Default Summary Prompt</label>
+          <textarea class="form-textarea" id="s-prompt" rows="4"></textarea>
+        </div>
+        <div class="btn-row">
+          <button class="btn btn-blue" onclick="saveSettings()">Save Defaults</button>
+        </div>
+      </div>
+
+      <div class="card">
+        <div class="card-header">
+          <div class="card-title">🔌 MCP Server Registration</div>
+          <div class="card-subtitle">Add this to your MCP settings to use with AI agents</div>
+        </div>
+        <pre id="mcp-config-snippet" style="background:var(--bg);border:1px solid var(--border);border-radius:4px;padding:14px;font-size:12px;color:var(--text-dim);overflow-x:auto;white-space:pre-wrap;word-break:break-all;"></pre>
+        <div class="mt-8">
+          <button class="btn btn-outline btn-sm" onclick="copyMcpConfig()">📋 Copy to Clipboard</button>
+        </div>
+      </div>
+    </div>
+
+  </main>
+</div>
+
+<!-- ═══════════════════════ FOLDER BROWSER MODAL ═══════════════════════ -->
+<div class="modal-overlay hidden" id="folder-modal">
+  <div class="modal" style="max-width:600px">
+    <div class="modal-title">📁 Choose Output Folder</div>
+    <div style="margin-bottom:10px;">
+      <div class="input-with-btn">
+        <input type="text" class="form-input" id="folder-modal-path" placeholder="Type path or browse below…" />
+        <button class="btn btn-blue btn-sm" onclick="navigateToTyped()">Go</button>
+        <button class="btn btn-outline btn-sm" onclick="createNewFolder()">+ New Folder</button>
+      </div>
+    </div>
+    <div class="folder-current" id="folder-current-path"></div>
+    <div class="folder-browser" id="folder-browser-list"></div>
+    <div class="btn-row mt-16">
+      <button class="btn btn-green" onclick="selectCurrentFolder()">✔ Select This Folder</button>
+      <button class="btn btn-outline" onclick="closeFolderModal()">Cancel</button>
+    </div>
+  </div>
+</div>
+
+<!-- ═══════════════════════ TOAST CONTAINER ═══════════════════════ -->
+<div class="toast-container" id="toast-container"></div>
+
+<script src="/app.js"></script>
+</body>
+</html>
diff --git a/backend/web/style.css b/backend/web/style.css
new file mode 100644
index 0000000..601e87b
--- /dev/null
+++ b/backend/web/style.css
@@ -0,0 +1,778 @@
+/* ============================================================
+   YouTube SIPOC Tool - Dark Mode UI
+   ============================================================ */
+
+:root {
+  --bg: #0f0f0f;
+  --bg2: #1a1a1a;
+  --bg3: #242424;
+  --bg4: #2e2e2e;
+  --border: #3a3a3a;
+  --text: #e8e8e8;
+  --text-dim: #888;
+  --text-muted: #555;
+  --accent: #ff4444;
+  --accent2: #cc2222;
+  --blue: #4a9eff;
+  --green: #22cc66;
+  --yellow: #ffcc00;
+  --orange: #ff8822;
+  --purple: #9966ff;
+  --red: #ff4444;
+  --radius: 8px;
+  --radius-sm: 4px;
+  --shadow: 0 4px 20px rgba(0,0,0,0.4);
+  --font: 'Segoe UI', system-ui, -apple-system, sans-serif;
+  --font-mono: 'Cascadia Code', 'Fira Code', 'Consolas', monospace;
+}
+
+* { box-sizing: border-box; margin: 0; padding: 0; }
+
+body {
+  font-family: var(--font);
+  background: var(--bg);
+  color: var(--text);
+  font-size: 14px;
+  line-height: 1.5;
+  min-height: 100vh;
+}
+
+/* ── Layout ───────────────────────────────────────────────── */
+
+.app-header {
+  background: var(--bg2);
+  border-bottom: 1px solid var(--border);
+  padding: 0 24px;
+  display: flex;
+  align-items: center;
+  gap: 16px;
+  height: 56px;
+  position: sticky;
+  top: 0;
+  z-index: 100;
+}
+
+.app-logo {
+  display: flex;
+  align-items: center;
+  gap: 10px;
+  text-decoration: none;
+}
+
+.app-logo svg { color: var(--accent); }
+
+.app-title {
+  font-size: 18px;
+  font-weight: 700;
+  color: var(--text);
+  letter-spacing: -0.3px;
+}
+
+.app-title span { color: var(--accent); }
+
+.header-nav {
+  display: flex;
+  gap: 4px;
+  margin-left: auto;
+}
+
+.nav-btn {
+  background: none;
+  border: 1px solid transparent;
+  color: var(--text-dim);
+  padding: 6px 14px;
+  border-radius: var(--radius-sm);
+  cursor: pointer;
+  font-size: 13px;
+  transition: all 0.15s;
+}
+.nav-btn:hover { background: var(--bg3); color: var(--text); border-color: var(--border); }
+.nav-btn.active { background: var(--bg3); color: var(--text); border-color: var(--border); }
+
+.app-body {
+  display: flex;
+  min-height: calc(100vh - 56px);
+}
+
+/* Sidebar */
+.sidebar {
+  width: 260px;
+  background: var(--bg2);
+  border-right: 1px solid var(--border);
+  display: flex;
+  flex-direction: column;
+  flex-shrink: 0;
+}
+
+.sidebar-section {
+  padding: 12px;
+  border-bottom: 1px solid var(--border);
+}
+
+.sidebar-label {
+  font-size: 10px;
+  font-weight: 700;
+  text-transform: uppercase;
+  letter-spacing: 0.8px;
+  color: var(--text-muted);
+  margin-bottom: 8px;
+  padding: 0 4px;
+}
+
+.job-list { display: flex; flex-direction: column; gap: 4px; }
+
+.job-item {
+  padding: 8px 10px;
+  border-radius: var(--radius-sm);
+  cursor: pointer;
+  transition: background 0.12s;
+  border: 1px solid transparent;
+}
+.job-item:hover { background: var(--bg3); }
+.job-item.active { background: var(--bg3); border-color: var(--border); }
+
+.job-item-title {
+  font-size: 12px;
+  font-weight: 600;
+  color: var(--text);
+  white-space: nowrap;
+  overflow: hidden;
+  text-overflow: ellipsis;
+}
+
+.job-item-meta {
+  font-size: 11px;
+  color: var(--text-dim);
+  display: flex;
+  align-items: center;
+  gap: 6px;
+  margin-top: 2px;
+}
+
+.new-job-btn {
+  width: 100%;
+  background: var(--accent);
+  color: #fff;
+  border: none;
+  padding: 9px;
+  border-radius: var(--radius-sm);
+  cursor: pointer;
+  font-size: 13px;
+  font-weight: 600;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  gap: 6px;
+  transition: background 0.15s;
+}
+.new-job-btn:hover { background: var(--accent2); }
+
+/* Main content */
+.main-content {
+  flex: 1;
+  overflow-y: auto;
+  padding: 24px;
+  max-width: 1100px;
+}
+
+/* ── Cards & Panels ──────────────────────────────────────── */
+
+.card {
+  background: var(--bg2);
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  padding: 20px;
+  margin-bottom: 16px;
+}
+
+.card-header {
+  display: flex;
+  align-items: center;
+  gap: 10px;
+  margin-bottom: 16px;
+  padding-bottom: 12px;
+  border-bottom: 1px solid var(--border);
+}
+
+.card-title {
+  font-size: 15px;
+  font-weight: 700;
+  color: var(--text);
+}
+
+.card-subtitle {
+  font-size: 12px;
+  color: var(--text-dim);
+  margin-top: 1px;
+}
+
+/* ── Form Controls ────────────────────────────────────────── */
+
+.form-group {
+  margin-bottom: 16px;
+}
+
+.form-label {
+  display: block;
+  font-size: 12px;
+  font-weight: 600;
+  color: var(--text-dim);
+  margin-bottom: 6px;
+  text-transform: uppercase;
+  letter-spacing: 0.4px;
+}
+
+.form-input,
+.form-select,
+.form-textarea {
+  width: 100%;
+  background: var(--bg3);
+  border: 1px solid var(--border);
+  color: var(--text);
+  padding: 8px 12px;
+  border-radius: var(--radius-sm);
+  font-family: var(--font);
+  font-size: 13px;
+  transition: border-color 0.15s;
+  outline: none;
+}
+.form-input:focus,
+.form-select:focus,
+.form-textarea:focus { border-color: var(--blue); }
+.form-textarea { resize: vertical; min-height: 80px; }
+.form-input.error { border-color: var(--red); }
+
+.input-with-btn {
+  display: flex;
+  gap: 8px;
+}
+.input-with-btn .form-input { flex: 1; }
+
+/* Type pills */
+.type-pills {
+  display: flex;
+  gap: 6px;
+  flex-wrap: wrap;
+}
+
+.type-pill {
+  padding: 5px 12px;
+  border-radius: 20px;
+  border: 1px solid var(--border);
+  background: var(--bg3);
+  color: var(--text-dim);
+  font-size: 12px;
+  cursor: pointer;
+  transition: all 0.15s;
+  user-select: none;
+}
+.type-pill:hover { border-color: var(--blue); color: var(--text); }
+.type-pill.selected { background: var(--blue); border-color: var(--blue); color: #fff; }
+
+/* Artifact checkboxes */
+.artifact-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fill, minmax(130px, 1fr));
+  gap: 8px;
+}
+
+.artifact-check {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  padding: 8px 12px;
+  background: var(--bg3);
+  border: 1px solid var(--border);
+  border-radius: var(--radius-sm);
+  cursor: pointer;
+  transition: all 0.15s;
+  user-select: none;
+}
+.artifact-check:hover { border-color: var(--blue); }
+.artifact-check.checked { border-color: var(--green); background: rgba(34,204,102,0.08); }
+.artifact-check input { display: none; }
+
+.artifact-icon { font-size: 16px; width: 20px; text-align: center; }
+.artifact-label { font-size: 12px; font-weight: 600; color: var(--text); }
+
+/* Range inputs */
+.range-row {
+  display: flex;
+  gap: 12px;
+}
+.range-row .form-group { flex: 1; }
+
+/* Folder picker */
+.folder-display {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  padding: 8px 12px;
+  background: var(--bg3);
+  border: 1px solid var(--border);
+  border-radius: var(--radius-sm);
+  font-size: 13px;
+  color: var(--text-dim);
+  font-family: var(--font-mono);
+  cursor: pointer;
+  transition: border-color 0.15s;
+  min-height: 36px;
+}
+.folder-display:hover { border-color: var(--blue); }
+.folder-display .folder-path { flex: 1; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; color: var(--text); }
+
+/* ── Buttons ──────────────────────────────────────────────── */
+
+.btn {
+  display: inline-flex;
+  align-items: center;
+  gap: 6px;
+  padding: 8px 16px;
+  border-radius: var(--radius-sm);
+  border: 1px solid transparent;
+  font-size: 13px;
+  font-weight: 600;
+  cursor: pointer;
+  transition: all 0.15s;
+  font-family: var(--font);
+}
+.btn:disabled { opacity: 0.4; cursor: not-allowed; }
+
+.btn-primary { background: var(--accent); color: #fff; }
+.btn-primary:hover:not(:disabled) { background: var(--accent2); }
+
+.btn-blue { background: var(--blue); color: #fff; }
+.btn-blue:hover:not(:disabled) { background: #3a8ee0; }
+
+.btn-green { background: var(--green); color: #000; }
+.btn-green:hover:not(:disabled) { background: #1ab055; }
+
+.btn-outline {
+  background: transparent;
+  border-color: var(--border);
+  color: var(--text-dim);
+}
+.btn-outline:hover:not(:disabled) { border-color: var(--text-dim); color: var(--text); }
+
+.btn-danger { background: rgba(255,68,68,0.15); border-color: var(--red); color: var(--red); }
+.btn-danger:hover:not(:disabled) { background: rgba(255,68,68,0.25); }
+
+.btn-sm { padding: 5px 10px; font-size: 12px; }
+
+.btn-row {
+  display: flex;
+  gap: 8px;
+  align-items: center;
+  flex-wrap: wrap;
+}
+
+/* ── Status Badges ────────────────────────────────────────── */
+
+.badge {
+  display: inline-flex;
+  align-items: center;
+  gap: 4px;
+  padding: 2px 8px;
+  border-radius: 20px;
+  font-size: 11px;
+  font-weight: 700;
+  text-transform: uppercase;
+  letter-spacing: 0.4px;
+}
+
+.badge-running { background: rgba(74,158,255,0.15); color: var(--blue); border: 1px solid rgba(74,158,255,0.3); }
+.badge-done { background: rgba(34,204,102,0.15); color: var(--green); border: 1px solid rgba(34,204,102,0.3); }
+.badge-done_with_errors { background: rgba(255,136,34,0.15); color: var(--orange); border: 1px solid rgba(255,136,34,0.3); }
+.badge-failed { background: rgba(255,68,68,0.15); color: var(--red); border: 1px solid rgba(255,68,68,0.3); }
+.badge-paused { background: rgba(255,204,0,0.15); color: var(--yellow); border: 1px solid rgba(255,204,0,0.3); }
+.badge-stopped { background: rgba(136,136,136,0.15); color: var(--text-dim); border: 1px solid var(--border); }
+.badge-pending { background: rgba(153,102,255,0.15); color: var(--purple); border: 1px solid rgba(153,102,255,0.3); }
+
+/* Artifact status */
+.art-dot {
+  display: inline-block;
+  width: 8px;
+  height: 8px;
+  border-radius: 50%;
+  flex-shrink: 0;
+}
+.art-dot-done { background: var(--green); }
+.art-dot-failed { background: var(--red); box-shadow: 0 0 4px var(--red); }
+.art-dot-running { background: var(--blue); animation: pulse 1s infinite; }
+.art-dot-pending { background: var(--border); }
+.art-dot-skipped { background: var(--text-muted); }
+
+@keyframes pulse {
+  0%, 100% { opacity: 1; }
+  50% { opacity: 0.4; }
+}
+
+/* ── Progress ─────────────────────────────────────────────── */
+
+.progress-bar {
+  width: 100%;
+  height: 6px;
+  background: var(--bg4);
+  border-radius: 3px;
+  overflow: hidden;
+  margin: 8px 0;
+}
+.progress-fill {
+  height: 100%;
+  border-radius: 3px;
+  background: var(--blue);
+  transition: width 0.4s ease;
+}
+.progress-fill.green { background: var(--green); }
+.progress-fill.yellow { background: var(--yellow); }
+.progress-fill.red { background: var(--red); }
+
+.stats-row {
+  display: flex;
+  gap: 16px;
+  flex-wrap: wrap;
+}
+
+.stat-item {
+  display: flex;
+  flex-direction: column;
+  gap: 2px;
+}
+.stat-value {
+  font-size: 22px;
+  font-weight: 700;
+  color: var(--text);
+  line-height: 1;
+}
+.stat-label {
+  font-size: 11px;
+  color: var(--text-muted);
+  text-transform: uppercase;
+  letter-spacing: 0.5px;
+}
+
+/* ── Video Table ──────────────────────────────────────────── */
+
+.video-table-wrap {
+  overflow-x: auto;
+  border-radius: var(--radius-sm);
+  border: 1px solid var(--border);
+}
+
+.video-table {
+  width: 100%;
+  border-collapse: collapse;
+  font-size: 13px;
+}
+
+.video-table th {
+  text-align: left;
+  padding: 10px 12px;
+  background: var(--bg3);
+  color: var(--text-muted);
+  font-size: 11px;
+  font-weight: 700;
+  text-transform: uppercase;
+  letter-spacing: 0.5px;
+  border-bottom: 1px solid var(--border);
+  white-space: nowrap;
+}
+
+.video-table td {
+  padding: 10px 12px;
+  border-bottom: 1px solid var(--border);
+  vertical-align: middle;
+}
+
+.video-table tr:last-child td { border-bottom: none; }
+.video-table tr:hover td { background: rgba(255,255,255,0.02); }
+.video-table tr.active-row td { background: rgba(74,158,255,0.06); }
+
+.video-title-cell {
+  max-width: 300px;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.video-id-badge {
+  font-family: var(--font-mono);
+  font-size: 11px;
+  color: var(--text-muted);
+  background: var(--bg3);
+  padding: 2px 6px;
+  border-radius: 3px;
+}
+
+.artifact-pills {
+  display: flex;
+  gap: 4px;
+  flex-wrap: wrap;
+}
+
+.art-pill {
+  display: inline-flex;
+  align-items: center;
+  gap: 3px;
+  padding: 2px 7px;
+  border-radius: 10px;
+  font-size: 11px;
+  font-weight: 600;
+  border: 1px solid var(--border);
+  background: var(--bg3);
+  color: var(--text-muted);
+  white-space: nowrap;
+}
+.art-pill-done { border-color: rgba(34,204,102,0.4); background: rgba(34,204,102,0.08); color: var(--green); }
+.art-pill-failed { border-color: rgba(255,68,68,0.4); background: rgba(255,68,68,0.08); color: var(--red); }
+.art-pill-running { border-color: rgba(74,158,255,0.4); background: rgba(74,158,255,0.08); color: var(--blue); }
+.art-pill-pending { color: var(--text-muted); }
+
+/* ── Log Panel ────────────────────────────────────────────── */
+
+.log-panel {
+  background: var(--bg);
+  border: 1px solid var(--border);
+  border-radius: var(--radius-sm);
+  height: 300px;
+  overflow-y: auto;
+  font-family: var(--font-mono);
+  font-size: 12px;
+  padding: 12px;
+}
+
+.log-entry {
+  display: flex;
+  gap: 10px;
+  padding: 2px 0;
+  line-height: 1.4;
+}
+
+.log-ts { color: var(--text-muted); flex-shrink: 0; }
+.log-level { flex-shrink: 0; font-weight: 700; min-width: 40px; }
+.log-level-INFO { color: var(--blue); }
+.log-level-WARN { color: var(--yellow); }
+.log-level-ERROR { color: var(--red); }
+.log-level-DEBUG { color: var(--text-muted); }
+.log-msg { color: var(--text-dim); word-break: break-all; }
+
+/* ── Modal ────────────────────────────────────────────────── */
+
+.modal-overlay {
+  position: fixed;
+  inset: 0;
+  background: rgba(0,0,0,0.7);
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  z-index: 999;
+  backdrop-filter: blur(4px);
+}
+
+.modal {
+  background: var(--bg2);
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  padding: 24px;
+  width: 90%;
+  max-width: 520px;
+  max-height: 80vh;
+  overflow-y: auto;
+  box-shadow: var(--shadow);
+}
+
+.modal-title {
+  font-size: 16px;
+  font-weight: 700;
+  margin-bottom: 16px;
+  color: var(--text);
+}
+
+/* Folder browser */
+.folder-browser {
+  background: var(--bg3);
+  border: 1px solid var(--border);
+  border-radius: var(--radius-sm);
+  max-height: 320px;
+  overflow-y: auto;
+}
+
+.folder-entry {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  padding: 8px 12px;
+  cursor: pointer;
+  border-bottom: 1px solid var(--border);
+  transition: background 0.1s;
+  font-size: 13px;
+}
+.folder-entry:last-child { border-bottom: none; }
+.folder-entry:hover { background: var(--bg4); }
+.folder-entry.is-file { color: var(--text-muted); cursor: default; }
+.folder-entry .entry-icon { width: 18px; text-align: center; font-size: 15px; }
+
+.folder-current {
+  font-family: var(--font-mono);
+  font-size: 11px;
+  color: var(--text-dim);
+  padding: 8px 12px;
+  border-bottom: 1px solid var(--border);
+  background: var(--bg4);
+  word-break: break-all;
+}
+
+/* ── Toast Notifications ─────────────────────────────────── */
+
+.toast-container {
+  position: fixed;
+  bottom: 24px;
+  right: 24px;
+  display: flex;
+  flex-direction: column;
+  gap: 8px;
+  z-index: 9999;
+}
+
+.toast {
+  background: var(--bg3);
+  border: 1px solid var(--border);
+  border-radius: var(--radius-sm);
+  padding: 12px 16px;
+  min-width: 260px;
+  max-width: 400px;
+  box-shadow: var(--shadow);
+  display: flex;
+  align-items: flex-start;
+  gap: 10px;
+  animation: slide-in 0.2s ease;
+}
+.toast.success { border-left: 3px solid var(--green); }
+.toast.error { border-left: 3px solid var(--red); }
+.toast.info { border-left: 3px solid var(--blue); }
+.toast-msg { font-size: 13px; color: var(--text); flex: 1; }
+
+@keyframes slide-in {
+  from { transform: translateX(20px); opacity: 0; }
+  to { transform: translateX(0); opacity: 1; }
+}
+
+/* ── Settings ─────────────────────────────────────────────── */
+
+.settings-grid {
+  display: grid;
+  grid-template-columns: 1fr 1fr;
+  gap: 12px;
+}
+@media (max-width: 600px) { .settings-grid { grid-template-columns: 1fr; } }
+
+/* ── Empty States ─────────────────────────────────────────── */
+
+.empty-state {
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+  justify-content: center;
+  padding: 60px 24px;
+  text-align: center;
+  gap: 12px;
+  color: var(--text-muted);
+}
+.empty-state-icon { font-size: 48px; }
+.empty-state-title { font-size: 16px; font-weight: 700; color: var(--text-dim); }
+.empty-state-body { font-size: 13px; max-width: 360px; }
+
+/* ── Misc ─────────────────────────────────────────────────── */
+
+.divider {
+  border: none;
+  border-top: 1px solid var(--border);
+  margin: 16px 0;
+}
+
+.section-title {
+  font-size: 13px;
+  font-weight: 700;
+  color: var(--text-dim);
+  margin-bottom: 12px;
+  text-transform: uppercase;
+  letter-spacing: 0.5px;
+}
+
+.text-muted { color: var(--text-muted); }
+.text-dim { color: var(--text-dim); }
+.text-green { color: var(--green); }
+.text-red { color: var(--red); }
+.text-blue { color: var(--blue); }
+.text-yellow { color: var(--yellow); }
+
+.monospace { font-family: var(--font-mono); }
+
+.flex-row { display: flex; align-items: center; gap: 8px; }
+.flex-1 { flex: 1; }
+.mt-8 { margin-top: 8px; }
+.mt-16 { margin-top: 16px; }
+.mb-8 { margin-bottom: 8px; }
+
+.hidden { display: none !important; }
+
+/* Spinner */
+.spinner {
+  display: inline-block;
+  width: 14px;
+  height: 14px;
+  border: 2px solid rgba(255,255,255,0.2);
+  border-top-color: #fff;
+  border-radius: 50%;
+  animation: spin 0.7s linear infinite;
+  flex-shrink: 0;
+}
+@keyframes spin { to { transform: rotate(360deg); } }
+
+/* ── Job Control Bar ──────────────────────────────────────── */
+
+.job-controls {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  padding: 12px 0;
+  flex-wrap: wrap;
+}
+
+.current-video-info {
+  flex: 1;
+  font-size: 12px;
+  color: var(--text-dim);
+  font-family: var(--font-mono);
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+/* Scrollbar */
+::-webkit-scrollbar { width: 6px; height: 6px; }
+::-webkit-scrollbar-track { background: var(--bg); }
+::-webkit-scrollbar-thumb { background: var(--bg4); border-radius: 3px; }
+::-webkit-scrollbar-thumb:hover { background: var(--border); }
+
+/* View tabs */
+.view-tabs {
+  display: flex;
+  gap: 0;
+  border-bottom: 1px solid var(--border);
+  margin-bottom: 16px;
+}
+.view-tab {
+  padding: 8px 16px;
+  background: none;
+  border: none;
+  color: var(--text-muted);
+  font-size: 13px;
+  font-weight: 600;
+  cursor: pointer;
+  border-bottom: 2px solid transparent;
+  margin-bottom: -1px;
+  transition: all 0.15s;
+}
+.view-tab:hover { color: var(--text); }
+.view-tab.active { color: var(--blue); border-bottom-color: var(--blue); }
diff --git a/backend/worker.py b/backend/worker.py
new file mode 100644
index 0000000..42d72f8
--- /dev/null
+++ b/backend/worker.py
@@ -0,0 +1,507 @@
+"""
+worker.py - Background job pipeline for YouTube SIPOC Tool
+Handles pause/stop/resume, per-artifact saving, duplicate skipping,
+random delays, retry logic, and resume-on-failure.
+"""
+import time
+import random
+import logging
+import threading
+from typing import Dict, Callable, Optional
+
+import database as db
+import extractors
+import config
+
+logger = logging.getLogger("worker")
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Job Control Registry
+# Stores threading.Event objects per job_id for pause/stop/skip signals.
+# ─────────────────────────────────────────────────────────────────────────────
+
+_job_controls: Dict[str, Dict] = {}
+_registry_lock = threading.Lock()
+
+
+def _get_controls(job_id: str) -> Dict:
+    with _registry_lock:
+        if job_id not in _job_controls:
+            _job_controls[job_id] = {
+                "pause": threading.Event(),   # set = paused
+                "stop": threading.Event(),    # set = stop requested
+                "skip": threading.Event(),    # set = skip current video
+            }
+        return _job_controls[job_id]
+
+
+def _cleanup_controls(job_id: str):
+    with _registry_lock:
+        _job_controls.pop(job_id, None)
+
+
+def pause_job(job_id: str):
+    _get_controls(job_id)["pause"].set()
+    db.update_job_status(job_id, "paused")
+    db.add_log(job_id, "Job paused by user.", level="INFO")
+
+
+def resume_job(job_id: str):
+    _get_controls(job_id)["pause"].clear()
+    db.update_job_status(job_id, "running")
+    db.add_log(job_id, "Job resumed by user.", level="INFO")
+
+
+def stop_job(job_id: str):
+    _get_controls(job_id)["stop"].set()
+    _get_controls(job_id)["pause"].clear()  # Unpause so thread can exit
+    db.update_job_status(job_id, "stopped")
+    db.add_log(job_id, "Job stopped by user.", level="INFO")
+
+
+def skip_video(job_id: str):
+    """Signal the worker to skip the current video and move to the next."""
+    _get_controls(job_id)["skip"].set()
+    db.add_log(job_id, "Skip requested for current video.", level="INFO")
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SSE Event Emitter
+# ─────────────────────────────────────────────────────────────────────────────
+
+class EventEmitter:
+    """Thread-safe event queue for Server-Sent Events."""
+
+    def __init__(self):
+        self._listeners: Dict[str, list] = {}
+        self._lock = threading.Lock()
+
+    def subscribe(self, job_id: str) -> "EventQueue":
+        queue = EventQueue()
+        with self._lock:
+            if job_id not in self._listeners:
+                self._listeners[job_id] = []
+            self._listeners[job_id].append(queue)
+        return queue
+
+    def unsubscribe(self, job_id: str, queue: "EventQueue"):
+        with self._lock:
+            if job_id in self._listeners:
+                try:
+                    self._listeners[job_id].remove(queue)
+                except ValueError:
+                    pass
+
+    def emit(self, job_id: str, event_type: str, data: dict):
+        with self._lock:
+            listeners = list(self._listeners.get(job_id, []))
+        for q in listeners:
+            q.put(event_type, data)
+
+
+class EventQueue:
+    """Per-subscriber event queue with thread-safe get."""
+
+    def __init__(self):
+        import queue
+        self._q = queue.Queue(maxsize=500)
+
+    def put(self, event_type: str, data: dict):
+        try:
+            self._q.put_nowait({"type": event_type, "data": data})
+        except Exception:
+            pass  # Drop events if queue is full
+
+    def get(self, timeout: float = 30.0):
+        try:
+            return self._q.get(timeout=timeout)
+        except Exception:
+            return None
+
+
+# Global emitter instance
+emitter = EventEmitter()
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Worker Thread Entry Point
+# ─────────────────────────────────────────────────────────────────────────────
+
+def run_job(job_id: str):
+    """
+    Main worker function. Runs in a background thread.
+    Resolves inputs, iterates videos, processes each artifact,
+    emits SSE events, handles pause/stop/skip/retry.
+    """
+    controls = _get_controls(job_id)
+    job = db.get_job(job_id)
+    if not job:
+        logger.error(f"Job {job_id} not found in database.")
+        return
+
+    db.update_job_status(job_id, "running")
+    _emit(job_id, "job_status", {"job_id": job_id, "status": "running"})
+    db.add_log(job_id, "Job started.", level="INFO")
+
+    try:
+        # ── Step 1: Resolve input to video list ──────────────────────────────
+        _emit(job_id, "progress", {"message": "Resolving input URLs…"})
+        db.add_log(job_id, f"Resolving input: type={job['input_type']}", level="INFO")
+
+        try:
+            videos = extractors.resolve_input(job["input_type"], job["input_value"])
+        except Exception as e:
+            db.add_log(job_id, f"Failed to resolve input: {e}", level="ERROR")
+            db.update_job_status(job_id, "failed")
+            _emit(job_id, "job_status", {"job_id": job_id, "status": "failed", "error": str(e)})
+            return
+
+        if not videos:
+            db.add_log(job_id, "No videos found for input.", level="WARN")
+            db.update_job_status(job_id, "done")
+            _emit(job_id, "job_status", {"job_id": job_id, "status": "done"})
+            return
+
+        # Deduplicate by video_id (preserve order)
+        seen = set()
+        unique_videos = []
+        for v in videos:
+            if v["video_id"] not in seen:
+                seen.add(v["video_id"])
+                unique_videos.append(v)
+        videos = unique_videos
+
+        # ── Step 2: Register videos in DB ────────────────────────────────────
+        for i, v in enumerate(videos):
+            db.upsert_video(job_id, v["video_id"], v["url"], v.get("title", ""), i)
+        db.update_job_progress(job_id, len(videos), 0)
+        _emit(job_id, "video_list", {
+            "job_id": job_id,
+            "total": len(videos),
+            "videos": [{"video_id": v["video_id"], "title": v.get("title", ""), "position": i}
+                       for i, v in enumerate(videos)]
+        })
+        db.add_log(job_id, f"Found {len(videos)} unique videos.", level="INFO")
+
+        # ── Step 3: Process each video ────────────────────────────────────────
+        selections = job["selections"]
+        done_count = 0
+
+        for idx, video in enumerate(videos):
+            # Check stop
+            if controls["stop"].is_set():
+                db.add_log(job_id, "Job stopped before processing video.", level="INFO")
+                break
+
+            # Handle pause (blocking)
+            _wait_if_paused(controls)
+            if controls["stop"].is_set():
+                break
+
+            # Reset skip flag for this video
+            controls["skip"].clear()
+
+            video_id = video["video_id"]
+            db.add_log(job_id, f"Processing video {idx+1}/{len(videos)}: {video_id}", level="INFO")
+            _emit(job_id, "video_start", {"job_id": job_id, "video_id": video_id, "position": idx})
+
+            # ── Step 3a: Fetch metadata (once per video) ──────────────────
+            meta = {}
+            try:
+                _emit(job_id, "artifact_start", {
+                    "job_id": job_id, "video_id": video_id, "artifact_type": "metadata"
+                })
+                meta = extractors.fetch_metadata(video_id)
+                title = meta.get("title", "")
+                if title:
+                    db.update_video_title(job_id, video_id, title)
+                    _emit(job_id, "video_title", {"job_id": job_id, "video_id": video_id, "title": title})
+                db.add_log(job_id, f"Metadata fetched for {video_id}: {title or '(no title)'}", level="INFO")
+            except Exception as e:
+                db.add_log(job_id, f"Failed to fetch metadata for {video_id}: {e}", level="ERROR")
+                _emit(job_id, "video_error", {"job_id": job_id, "video_id": video_id, "error": str(e)})
+                db.update_video_status(job_id, video_id, "failed")
+                continue
+
+            # ── Step 3b: Initialize artifact records ──────────────────────
+            for artifact_type in selections:
+                db.upsert_artifact(job_id, video_id, artifact_type)
+
+            # ── Step 3c: Process each selected artifact ───────────────────
+            video_failed = False
+            for artifact_type in selections:
+                # Check stop or skip
+                if controls["stop"].is_set():
+                    break
+                if controls["skip"].is_set():
+                    db.add_log(job_id, f"Skipping remaining artifacts for {video_id} (skip requested).", level="INFO")
+                    break
+
+                # Handle pause
+                _wait_if_paused(controls)
+                if controls["stop"].is_set():
+                    break
+
+                # Skip if already done
+                if db.artifact_is_done(job_id, video_id, artifact_type):
+                    db.add_log(job_id, f"[SKIP-DUP] {artifact_type} for {video_id} already done.", level="INFO")
+                    _emit(job_id, "artifact_skip", {
+                        "job_id": job_id, "video_id": video_id, "artifact_type": artifact_type,
+                        "reason": "already_done"
+                    })
+                    continue
+
+                # Process with retry
+                success = _process_artifact_with_retry(
+                    job_id=job_id,
+                    video_id=video_id,
+                    artifact_type=artifact_type,
+                    meta=meta,
+                    job=job,
+                    controls=controls,
+                )
+                if not success:
+                    video_failed = True
+
+                # Random delay between artifacts (bot avoidance)
+                if not controls["stop"].is_set() and not controls["skip"].is_set():
+                    delay = random.uniform(job["delay_min"], job["delay_max"])
+                    db.add_log(job_id, f"Waiting {delay:.1f}s before next artifact…", level="DEBUG")
+                    _interruptible_sleep(delay, controls)
+
+            # ── Step 3d: Mark video done/failed ──────────────────────────
+            if not controls["stop"].is_set():
+                status = "failed" if video_failed else "done"
+                db.update_video_status(job_id, video_id, status)
+                done_count += 1
+                db.update_job_progress(job_id, len(videos), done_count)
+                _emit(job_id, "video_done", {
+                    "job_id": job_id,
+                    "video_id": video_id,
+                    "status": status,
+                    "done": done_count,
+                    "total": len(videos),
+                })
+
+            # Random delay between videos
+            if idx < len(videos) - 1 and not controls["stop"].is_set():
+                delay = random.uniform(job["delay_min"], job["delay_max"])
+                db.add_log(job_id, f"Waiting {delay:.1f}s before next video…", level="DEBUG")
+                _interruptible_sleep(delay, controls)
+
+        # ── Step 4: Finalize job ──────────────────────────────────────────────
+        if controls["stop"].is_set():
+            final_status = "stopped"
+        else:
+            stats = db.get_job_stats(job_id)
+            final_status = "done" if stats["artifact_failed"] == 0 else "done_with_errors"
+
+        db.update_job_status(job_id, final_status)
+        stats = db.get_job_stats(job_id)
+        db.add_log(job_id, f"Job finished with status: {final_status}. Stats: {stats}", level="INFO")
+        _emit(job_id, "job_status", {"job_id": job_id, "status": final_status, "stats": stats})
+
+    except Exception as e:
+        logger.exception(f"Unexpected error in job {job_id}: {e}")
+        db.add_log(job_id, f"Unexpected error: {e}", level="ERROR")
+        db.update_job_status(job_id, "failed")
+        _emit(job_id, "job_status", {"job_id": job_id, "status": "failed", "error": str(e)})
+    finally:
+        _cleanup_controls(job_id)
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Artifact Processing with Retry
+# ─────────────────────────────────────────────────────────────────────────────
+
+def _process_artifact_with_retry(
+    job_id: str,
+    video_id: str,
+    artifact_type: str,
+    meta: Dict,
+    job: Dict,
+    controls: Dict,
+) -> bool:
+    """
+    Attempt to process one artifact, retrying up to MAX_RETRIES times.
+    Returns True on success, False on final failure.
+    """
+    max_retries = config.MAX_RETRIES
+    output_folder = job["output_folder"]
+    resolution = job.get("resolution", "best")
+    api_key = job.get("openrouter_model", "")  # model field
+    # We store the API key in the job separately; retrieve from job dict
+    # (api_key is set in the job via the POST body - see api_server.py)
+    openrouter_api_key = job.get("_openrouter_api_key", "")  # injected by api_server
+    openrouter_model = job.get("openrouter_model", config.OPENROUTER_MODEL)
+    summary_prompt = job.get("summary_prompt", config.SUMMARY_SYSTEM_PROMPT)
+
+    for attempt in range(1, max_retries + 1):
+        if controls["stop"].is_set() or controls["skip"].is_set():
+            return False
+
+        db.mark_artifact_started(job_id, video_id, artifact_type)
+        _emit(job_id, "artifact_start", {
+            "job_id": job_id,
+            "video_id": video_id,
+            "artifact_type": artifact_type,
+            "attempt": attempt,
+        })
+        db.add_log(job_id, f"[{artifact_type}] Starting (attempt {attempt}/{max_retries})", level="INFO",
+                   video_id=video_id, artifact_type=artifact_type)
+
+        try:
+            file_path = _run_extractor(
+                artifact_type=artifact_type,
+                video_id=video_id,
+                meta=meta,
+                output_folder=output_folder,
+                resolution=resolution,
+                openrouter_api_key=openrouter_api_key,
+                openrouter_model=openrouter_model,
+                summary_prompt=summary_prompt,
+            )
+            db.update_artifact_status(job_id, video_id, artifact_type, "done", file_path=file_path)
+            db.add_log(job_id, f"[{artifact_type}] Done → {file_path}", level="INFO",
+                       video_id=video_id, artifact_type=artifact_type)
+            _emit(job_id, "artifact_done", {
+                "job_id": job_id,
+                "video_id": video_id,
+                "artifact_type": artifact_type,
+                "file_path": file_path,
+            })
+            return True
+
+        except Exception as e:
+            err_msg = str(e)
+            logger.warning(f"[{job_id}][{video_id}][{artifact_type}] Attempt {attempt} failed: {e}")
+            db.add_log(job_id, f"[{artifact_type}] Attempt {attempt} failed: {e}", level="WARN",
+                       video_id=video_id, artifact_type=artifact_type)
+
+            if attempt < max_retries:
+                # Exponential backoff between retries
+                backoff = min(2 ** attempt * 3, 30) + random.uniform(0, 2)
+                db.add_log(job_id, f"[{artifact_type}] Retry in {backoff:.1f}s…", level="INFO",
+                           video_id=video_id, artifact_type=artifact_type)
+                _interruptible_sleep(backoff, controls)
+            else:
+                db.update_artifact_status(
+                    job_id, video_id, artifact_type, "failed", error_msg=err_msg
+                )
+                db.add_log(job_id, f"[{artifact_type}] FAILED after {max_retries} attempts: {e}", level="ERROR",
+                           video_id=video_id, artifact_type=artifact_type)
+                _emit(job_id, "artifact_failed", {
+                    "job_id": job_id,
+                    "video_id": video_id,
+                    "artifact_type": artifact_type,
+                    "error": err_msg,
+                    "attempts": max_retries,
+                })
+                return False
+
+    return False
+
+
+def _run_extractor(
+    artifact_type: str,
+    video_id: str,
+    meta: Dict,
+    output_folder: str,
+    resolution: str,
+    openrouter_api_key: str,
+    openrouter_model: str,
+    summary_prompt: str,
+) -> str:
+    """Dispatch to the correct extractor function."""
+    if artifact_type == "title":
+        file_path, _ = extractors.extract_title(video_id, meta, output_folder)
+        return file_path
+    elif artifact_type == "stats":
+        return extractors.extract_stats(video_id, meta, output_folder)
+    elif artifact_type == "transcript":
+        return extractors.extract_transcript(video_id, meta, output_folder)
+    elif artifact_type == "summary":
+        return extractors.extract_summary(
+            video_id, meta, output_folder,
+            api_key=openrouter_api_key,
+            model=openrouter_model,
+            system_prompt=summary_prompt,
+        )
+    elif artifact_type == "thumbnail":
+        return extractors.extract_thumbnail(video_id, meta, output_folder)
+    elif artifact_type == "mp3":
+        return extractors.extract_mp3(video_id, meta, output_folder)
+    elif artifact_type == "mp4":
+        return extractors.extract_mp4(video_id, meta, output_folder, resolution=resolution)
+    else:
+        raise ValueError(f"Unknown artifact type: {artifact_type}")
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Helpers
+# ─────────────────────────────────────────────────────────────────────────────
+
+def _emit(job_id: str, event_type: str, data: dict):
+    """Emit an SSE event."""
+    emitter.emit(job_id, event_type, data)
+
+
+def _wait_if_paused(controls: Dict):
+    """Block worker thread while paused, check every 0.5s."""
+    while controls["pause"].is_set() and not controls["stop"].is_set():
+        time.sleep(0.5)
+
+
+def _interruptible_sleep(seconds: float, controls: Dict):
+    """Sleep for `seconds` but wake early if stop/skip is set."""
+    end = time.monotonic() + seconds
+    while time.monotonic() < end:
+        if controls["stop"].is_set() or controls["skip"].is_set():
+            return
+        time.sleep(min(0.3, end - time.monotonic()))
+
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Job Launcher
+# ─────────────────────────────────────────────────────────────────────────────
+
+def launch_job(job_id: str, openrouter_api_key: str = ""):
+    """
+    Launch job in a daemon background thread.
+    openrouter_api_key is injected at runtime (not stored in DB for security).
+    """
+    # Inject API key into job dict via a side channel stored in controls
+    controls = _get_controls(job_id)
+    controls["_openrouter_api_key"] = openrouter_api_key
+
+    # Patch the run_job function to use the injected key
+    def _patched_run():
+        job = db.get_job(job_id)
+        if job:
+            job["_openrouter_api_key"] = openrouter_api_key
+        run_job_with_key(job_id, openrouter_api_key)
+
+    t = threading.Thread(target=_patched_run, daemon=True, name=f"job-{job_id[:8]}")
+    t.start()
+    return t
+
+
+def run_job_with_key(job_id: str, openrouter_api_key: str):
+    """Wrapper that injects the API key into job processing."""
+    # Monkey-patch the job dict to include the API key during this run
+    original_get_job = db.get_job
+
+    def patched_get_job(jid):
+        result = original_get_job(jid)
+        if result and jid == job_id:
+            result["_openrouter_api_key"] = openrouter_api_key
+        return result
+
+    # Temporarily patch db.get_job for this thread
+    import database
+    _orig = database.get_job
+    database.get_job = patched_get_job
+    try:
+        run_job(job_id)
+    finally:
+        database.get_job = _orig
diff --git a/backend/youtube_sipoc.db b/backend/youtube_sipoc.db
new file mode 100644
index 0000000..6e1c6b7
Binary files /dev/null and b/backend/youtube_sipoc.db differ
diff --git a/loading.html b/loading.html
new file mode 100644
index 0000000..5fbfa24
--- /dev/null
+++ b/loading.html
@@ -0,0 +1,51 @@
+<!DOCTYPE html>
+<html>
+<head>
+<meta charset="UTF-8">
+<title>BatchTube — Starting...</title>
+<style>
+* { margin: 0; padding: 0; box-sizing: border-box; }
+body {
+  background: #0d0f14;
+  color: #e8eaf0;
+  font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+  justify-content: center;
+  height: 100vh;
+  gap: 24px;
+}
+.logo { font-size: 2.5rem; font-weight: 800; }
+.logo span { color: #FF8822; }
+.spinner {
+  width: 40px; height: 40px;
+  border: 3px solid rgba(255,136,34,0.2);
+  border-top-color: #FF8822;
+  border-radius: 50%;
+  animation: spin 0.8s linear infinite;
+}
+@keyframes spin { to { transform: rotate(360deg); } }
+.status { color: #8892a4; font-size: 0.9rem; }
+</style>
+</head>
+<body>
+<div class="logo">Batch<span>Tube</span></div>
+<div class="spinner"></div>
+<div class="status" id="status">Starting engine...</div>
+<script>
+async function checkServer() {
+  try {
+    const r = await fetch('http://localhost:8765/api/health');
+    if (r.ok) {
+      document.getElementById('status').textContent = 'Ready! Loading...'
+      window.location.href = 'http://localhost:8765';
+      return;
+    }
+  } catch(e) {}
+  setTimeout(checkServer, 500);
+}
+checkServer();
+</script>
+</body>
+</html>
diff --git a/package.json b/package.json
new file mode 100644
index 0000000..412fba9
--- /dev/null
+++ b/package.json
@@ -0,0 +1,16 @@
+{
+  "name": "batchtube",
+  "version": "1.0.0",
+  "description": "BatchTube — Batch YouTube Downloader",
+  "scripts": {
+    "tauri": "tauri",
+    "dev": "tauri dev",
+    "build": "tauri build"
+  },
+  "devDependencies": {
+    "@tauri-apps/cli": "^2.0.0"
+  },
+  "dependencies": {
+    "@tauri-apps/api": "^2.0.0"
+  }
+}
diff --git a/rollback.sh b/rollback.sh
new file mode 100644
index 0000000..37f134a
--- /dev/null
+++ b/rollback.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+# BatchTube Tauri Rollback Script
+# Run this if Tauri build fails and you need to revert to NSIS installer
+
+echo "=== BatchTube Rollback ==="
+echo "Reverting to pre-Tauri state..."
+
+# Check if rollback tag exists
+if git tag | grep -q "v1.0-pre-tauri"; then
+  git checkout v1.0-pre-tauri
+  echo "✅ Reverted to v1.0-pre-tauri"
+  echo "The original NSIS installer build is at: BatchTube App/BatchTube-Windows-Build.zip"
+  echo "To rebuild: run build_windows.bat"
+else
+  echo "❌ Rollback tag not found. Manual recovery needed."
+  echo "Original source: Batchtube-snapshot-030526.zip"
+fi
diff --git a/src-tauri/Cargo.toml b/src-tauri/Cargo.toml
new file mode 100644
index 0000000..b348ba0
--- /dev/null
+++ b/src-tauri/Cargo.toml
@@ -0,0 +1,25 @@
+[package]
+name = "batchtube"
+version = "1.0.0"
+edition = "2021"
+
+[lib]
+name = "batchtube_lib"
+crate-type = ["staticlib", "cdylib", "rlib"]
+
+[build-dependencies]
+tauri-build = { version = "2", features = [] }
+
+[dependencies]
+tauri = { version = "2", features = [] }
+tauri-plugin-shell = "2"
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+tokio = { version = "1", features = ["full"] }
+
+[profile.release]
+panic = "abort"
+codegen-units = 1
+lto = true
+opt-level = "s"
+strip = true
diff --git a/src-tauri/build.rs b/src-tauri/build.rs
new file mode 100644
index 0000000..795b9b7
--- /dev/null
+++ b/src-tauri/build.rs
@@ -0,0 +1,3 @@
+fn main() {
+  tauri_build::build()
+}
diff --git a/src-tauri/icons/128x128.png b/src-tauri/icons/128x128.png
new file mode 100644
index 0000000..59ec9c2
Binary files /dev/null and b/src-tauri/icons/128x128.png differ
diff --git a/src-tauri/icons/128x128@2x.png b/src-tauri/icons/128x128@2x.png
new file mode 100644
index 0000000..276cbd0
Binary files /dev/null and b/src-tauri/icons/128x128@2x.png differ
diff --git a/src-tauri/icons/32x32.png b/src-tauri/icons/32x32.png
new file mode 100644
index 0000000..59ec9c2
Binary files /dev/null and b/src-tauri/icons/32x32.png differ
diff --git a/src-tauri/icons/icon.ico b/src-tauri/icons/icon.ico
new file mode 100644
index 0000000..ebac682
Binary files /dev/null and b/src-tauri/icons/icon.ico differ
diff --git a/src-tauri/src/lib.rs b/src-tauri/src/lib.rs
new file mode 100644
index 0000000..6d521dc
--- /dev/null
+++ b/src-tauri/src/lib.rs
@@ -0,0 +1,8 @@
+#[cfg_attr(mobile, tauri::mobile_entry_point)]
+pub fn run() {
+    tauri::Builder::default()
+        .plugin(tauri_plugin_shell::init())
+        .invoke_handler(tauri::generate_handler![])
+        .run(tauri::generate_context!())
+        .expect("error while running tauri application");
+}
diff --git a/src-tauri/src/main.rs b/src-tauri/src/main.rs
new file mode 100644
index 0000000..446d502
--- /dev/null
+++ b/src-tauri/src/main.rs
@@ -0,0 +1,73 @@
+// Prevents additional console window on Windows in release
+#![cfg_attr(not(debug_assertions), windows_subsystem = "windows")]
+
+use std::sync::Mutex;
+use tauri::{Manager, State};
+use tauri_plugin_shell::ShellExt;
+
+struct ServerProcess(Mutex<Option<tauri_plugin_shell::process::CommandChild>>);
+
+#[tauri::command]
+async fn get_server_status() -> String {
+    match reqwest::get("http://localhost:8765/api/health").await {
+        Ok(resp) if resp.status().is_success() => "ready".to_string(),
+        _ => "starting".to_string(),
+    }
+}
+
+fn main() {
+    tauri::Builder::default()
+        .plugin(tauri_plugin_shell::init())
+        .setup(|app| {
+            let handle = app.handle().clone();
+
+            // Spawn the Python backend sidecar
+            let sidecar_command = app.shell().sidecar("batchtube-server").unwrap();
+            let (mut _rx, child) = sidecar_command
+                .spawn()
+                .expect("Failed to spawn BatchTube server");
+
+            // Store process handle for cleanup
+            app.manage(ServerProcess(Mutex::new(Some(child))));
+
+            // Wait for server to be ready, then show window
+            tauri::async_runtime::spawn(async move {
+                let mut attempts = 0;
+                loop {
+                    tokio::time::sleep(tokio::time::Duration::from_millis(500)).await;
+                    if let Ok(resp) = reqwest::get("http://localhost:8765/api/health").await {
+                        if resp.status().is_success() {
+                            // Server ready — show the main window
+                            if let Some(window) = handle.get_webview_window("main") {
+                                window.show().unwrap();
+                                window.set_focus().unwrap();
+                            }
+                            break;
+                        }
+                    }
+                    attempts += 1;
+                    if attempts > 30 {
+                        eprintln!("BatchTube server failed to start after 15 seconds");
+                        break;
+                    }
+                }
+            });
+
+            Ok(())
+        })
+        .on_window_event(|window, event| {
+            if let tauri::WindowEvent::CloseRequested { .. } = event {
+                // Kill the backend server on window close
+                if let Some(server) = window.app_handle().try_state::<ServerProcess>() {
+                    if let Ok(mut guard) = server.0.lock() {
+                        if let Some(child) = guard.take() {
+                            let _ = child.kill();
+                        }
+                    }
+                }
+            }
+        })
+        .invoke_handler(tauri::generate_handler![get_server_status])
+        .run(tauri::generate_context!())
+        .expect("error while running BatchTube");
+}
diff --git a/src-tauri/tauri.conf.json b/src-tauri/tauri.conf.json
new file mode 100644
index 0000000..96e9ff2
--- /dev/null
+++ b/src-tauri/tauri.conf.json
@@ -0,0 +1,50 @@
+{
+  "productName": "BatchTube",
+  "version": "1.0.0",
+  "identifier": "com.intentures.batchtube",
+  "build": {
+    "beforeDevCommand": "",
+    "beforeBuildCommand": "",
+    "devUrl": "http://localhost:8765",
+    "frontendDist": "http://localhost:8765"
+  },
+  "app": {
+    "windows": [
+      {
+        "title": "BatchTube",
+        "width": 1200,
+        "height": 800,
+        "minWidth": 900,
+        "minHeight": 600,
+        "resizable": true,
+        "fullscreen": false,
+        "center": true
+      }
+    ],
+    "security": {
+      "csp": null
+    }
+  },
+  "bundle": {
+    "active": true,
+    "targets": "all",
+    "icon": [
+      "icons/32x32.png",
+      "icons/128x128.png",
+      "icons/128x128@2x.png",
+      "icons/icon.icns",
+      "icons/icon.ico"
+    ],
+    "resources": [
+      "../backend/**/*"
+    ],
+    "externalBin": [
+      "../backend/batchtube-server"
+    ],
+    "windows": {
+      "certificateThumbprint": null,
+      "digestAlgorithm": "sha256",
+      "timestampUrl": ""
+    }
+  }
+}