- Premiere Pro 23.0.0 or higher
- Node.js 16+ and npm
- Python 3.8+
- UXP Developer Tools
- Google Gemini API Key (free tier available)
- Go to https://ai.google.dev/
- Click "Get API Key" → "Create API key in new project"
- Copy your API key (starts with
AIza...)
# Navigate to backend directory
cd backend
# Install Python dependencies
pip install -r requirements.txt
# Create .env file with your API key
echo "GEMINI_API_KEY=your_api_key_here" > .env
# On Windows PowerShell:
# New-Item -Path .env -ItemType File -Value "GEMINI_API_KEY=your_api_key_here"
# Start the backend server
python main.pyYou should see:
============================================================
ChatCut Backend Starting...
============================================================
Server: http://localhost:3001
Docs: http://localhost:3001/docs
Health: http://localhost:3001/health
============================================================
✓ Gemini service initialized successfully
Leave this terminal running!
Open a new terminal:
# Navigate to frontend directory
cd frontend
# Install dependencies (if you haven't already)
npm install
# Build the plugin
npm run watchYou should see:
webpack 4.47.0
Built at: ...
Asset Size
index.html 199 bytes
index.js 2.5 MiB
Leave this terminal running! It will auto-rebuild when you change code.
-
Launch Premiere Pro (version 23.0.0+)
-
Open UXP Developer Tools
-
Add the Plugin:
- Click "Add Plugin..."
- Navigate to:
D:\ChatCut\frontend\dist\manifest.json - Select it and click "Open"
-
Load the Plugin:
- Find "ChatCut" in the plugin list
- Click the ••• button → "Load"
-
Open the Panel in Premiere Pro:
- In Premiere Pro, go to: Window → Extensions → ChatCut
- The panel should appear!
- Look at the bottom of the ChatCut panel
- You should see:
Backend: ✓ Connected - If not, make sure the Python server is running on localhost:3001
- In Premiere Pro, open a sequence with video clips
- Set In and Out points on your timeline (press
IandO) - In the ChatCut panel, click "Get Current Selection"
- You should see the duration displayed
-
In the ChatCut panel, type a prompt:
- "make this black and white"
- "blur this clip"
- "zoom in 150%"
-
Click "Process Prompt"
-
You should see:
- AI Interpretation card with effect details
- Confidence score
- Description of what will happen
-
(Effect application is coming in Phase 3)
Error: GEMINI_API_KEY not found
- Make sure you created the
.envfile in thebackenddirectory - Check that the file contains:
GEMINI_API_KEY=your_actual_key
Error: ModuleNotFoundError: No module named 'google'
- Run:
pip install google-genai
Error: Invalid Version: npm:acorn-with-stage3
- Delete
package-lock.jsonandnode_modules - Run
npm installagain - (We already fixed this earlier!)
Error: "No applications are connected"
- Make sure Premiere Pro is running
- Restart UXP Developer Tools
- Try unloading and reloading the plugin
Panel doesn't appear
- In Premiere Pro: Window → Extensions → ChatCut
- Check UXP Developer Tools for error messages
Error: "Backend not connected" banner in panel
- Make sure Python server is running on port 3001
- Check firewall settings
- Try accessing http://localhost:3001/health in a browser
- Edit files in
frontend/src/ - Webpack will auto-rebuild (if
npm run watchis running) - In UXP Developer Tools, click ••• → "Reload"
- Panel updates automatically!
- Edit files in
backend/ - Stop the server (Ctrl+C)
- Restart:
python main.py - No need to reload the frontend
Frontend:
- In UXP Developer Tools, click ••• → "Debug"
- Opens Chrome DevTools
- Check Console for errors
Backend:
- Watch the terminal where
python main.pyis running - All requests are logged there
Once running, you can test the backend directly:
- Health Check: http://localhost:3001/health
- API Docs: http://localhost:3001/docs (interactive Swagger UI)
- Root: http://localhost:3001/
# Test prompt processing (requires Gemini API key)
curl -X POST http://localhost:3001/api/process-prompt \
-H "Content-Type: application/json" \
-d '{"prompt": "make this black and white"}'✅ Frontend panel loads in Premiere Pro ✅ Timeline selection detection ✅ Backend API connection ✅ Gemini AI prompt processing ✅ Effect mapping (Black & White, Blur, Transform) ✅ Confidence scoring ✅ Error handling
❌ Actually applying effects to timeline (Phase 3) ❌ Multi-track support ❌ Undo functionality ❌ AI video generation (out of scope for MVP)
See plan.md for the full roadmap!
Current Phase: Phase 2 Complete ✓ Next Phase: Phase 3 - Effect Application
- Check plan.md for architecture details
- See backend logs for API errors
- Use UXP Debug console for frontend errors
- Gemini API docs: https://ai.google.dev/gemini-api/docs
Ready to ship this MVP! 🚀