VidlyAI

VidlyAI is your intelligent companion for YouTube. It automatically fetches video transcripts and allows you to chat with any video using state-of-the-art AI models. Ask questions, get summaries, and navigate directly to the most relevant parts of the video.

Features

• 🤖 Multi-Provider AI Support:
  • Google Gemini (Gemini 2.5 Flash Lite)
  • OpenAI (GPT-4o Mini)
  • OpenRouter (DeepSeek, Llama, etc.)
  • Anthropic Claude (Haiku)
• 📝 Automatic Transcripts: Fetches transcripts automatically for the current video.
• ⚡️ Smart Context: The AI answers questions strictly based on the video context.
• 📍 Timestamp Navigation: Clickable timestamps in the chat to jump directly to that moment in the video.
• 🔐 Secure & Private: Your API keys are stored locally in your browser (Chrome Storage) and never sent to our servers.
• 🎨 Sleek UI: A modern, dark-mode interface that fits right into YouTube.

Installation (Developer Mode)

1. Clone the repository:

   git clone https://github.com/yourusername/VidlyAI.git
   cd VidlyAI

2. Install dependencies:

   npm install

3. Build the extension:

   npm run build

4. Load into Chrome:

   • Open chrome://extensions/
   • Enable Developer mode (top right).
   • Click Load unpacked.
   • Select the dist folder generated in this project.

Usage

1. Configure API Key:

   • Click the extension icon or open the Options Page.
   • Select your preferred AI provider (e.g., Gemini, OpenAI).
   • Enter your API Key.
   • Click Save.

2. Chat with a Video:

   • Open any YouTube video.
   • The VidlyAI sidebar will appear.
   • Wait for the "Context Active" badge.
   • Ask anything! (e.g., "Summarize this video", "What did they say about X?").

Tech Stack

• Frontend: React, TypeScript, Tailwind CSS
• Build Tool: Vite
• Bundling: Custom Vite config for Chrome Extensions (Content Scripts, Background, Options)
• Icons: Lucide React

License

MIT