Run and deploy your AI Studio app

This contains everything you need to run your app locally.

View your app in AI Studio: https://ai.studio/apps/drive/1ARYaFdx2arW9joHVDoUfDS06cWXXwmXp

Run Locally

Prerequisites: Node.js

1. Install dependencies: npm install
2. Create an .env.local file and set any Ollama defaults you want to preload:

   OLLAMA_URL="https://ollama.com/api"
   # Leave this unset to keep the default cloud vision model
   OLLAMA_MODEL="qwen3-vl:235b-instruct-cloud"
   OLLAMA_API_KEY="your-shared-ollama-key"
   

3. Start the local Ollama proxy (new terminal): npm run dev:proxy
4. Run the app: npm run dev

Photo + Video uploads

• Upload standard photos (JPG/PNG) or short videos (MP4/MOV/WebM) directly from your device.
• Videos are limited to ~45 seconds and ~80MB to keep frame extraction responsive during local analysis.
• When a video is uploaded or recorded from your camera, the app samples a handful of frames in chronological order and sends them to the selected vision model, so you can reuse the same Analyze flow for both media types.

Cloudflare Pages Functions proxy

• The app calls /api/ollama/generate and /api/ollama/test, which are implemented under functions/api/ollama/ for deployment on Cloudflare Pages Functions.
• Provide OLLAMA_URL, OLLAMA_MODEL (optional), and OLLAMA_API_KEY as environment variables locally (.env.local) and in your Pages project settings so the proxy can reach Ollama Cloud.
• Users can still supply their own URL/key via the Settings modal; the proxy simply forwards those credentials server-side to avoid browser CORS blocks.
• To store Ollama API keys securely after a successful connection test, bind a KV namespace named KEY_STORE (or update the binding name in code) to your Pages project. The /api/keys/ollama route writes or deletes the key in that namespace.
• Clearing keys: curl -X DELETE https://<your-domain>/api/keys/ollama