Alan

Alan continuously captures screenshots and performs image recognition. The results of the image recognition are logged for further analysis.

👋 Hey! Watch me code it in this 📺recorded livestream.

Features

• Automated screenshot capture
• Image recognition using Google's Gemini 2.0 Flash model
• Continuous results logging with timestamps
• Built with TypeScript and Bun runtime

Prerequisites

• Bun runtime installed
• Gemini API key
• TypeScript 5.x

Installation

1. Clone the repository:

   git clone https://github.com/gavmor/alan.git
   cd alan

2. Install dependencies:

   bun install

3. Set up your Gemini API key:

   • Visit Google AI Studio
   • Create an API key
   • Set the API key in your environment:

     export GEMINI_API_KEY='your-api-key-here'

Usage

Run the application:

bun run index.ts

The application will:

1. Take screenshots of your desktop
2. Process the screenshots through Gemini Pro Vision for image recognition
3. Append the recognition results with timestamps to results.txt

Development

To run tests:

bun test

Dependencies

• screenshot-desktop: For capturing desktop screenshots
  • (Linux) image-magick
• ollama: For interfacing with Ollama AI models
• TypeScript: For type-safe development