Doggo 🐕, your loyal digital companion who finds files the way you think about them.

---

Doggo is a CLI tool that uses AI to help you search for and organize images using natural language queries. Instead of remembering exact filenames, just describe what you're looking for!

Features

• 🔍 Semantic Search: Find images by describing them in natural language
• 🗂️ AI-Powered Organization: Automatically organize images into intelligent categories
• ✏️ Smart Renaming: Generate descriptive filenames based on image content
• 🎯 Smart Results: AI-powered similarity matching
• 💻 CLI Interface: Simple command-line interface
• 📊 Rich Output: Beautiful, informative search results

Demo

demo.mov

Usage

pip install doggo

Quick Start

1. Initialize Doggo:

   doggo init

2. Configure your AI provider:

   For OpenAI (default):

   doggo config set --api-key "sk-your-openai-api-key"

   For Ollama (local models):

   doggo config set --provider-url "http://localhost:11434/v1" \
                    --chat-model "gemma3:latest" \
                    --embedding-model "nomic-embed-text:latest"

   Custom OpenAI setup:

   doggo config set --provider-url "https://api.openai.com/v1" \
                    --chat-model "gpt-4" \
                    --embedding-model "text-embedding-ada-002" \
                    --api-key "sk-your-openai-api-key"

3. View current configuration:

   doggo config show

4. Index your images:

   doggo index /path/to/your/images

5. Search naturally:

   doggo search "a cute dog playing in the park"
   doggo search "sunset over mountains"
   doggo search "people having dinner"

   By default, Doggo shows the top 5 results and automatically opens the best match in your system's previewer. Use --no-open to disable auto-opening or --limit to change the number of results.

6. Organize your images:

   # Organize images into AI-generated category folders
   doggo organize /path/to/your/images
   
   # Organize with descriptive filenames
   doggo organize /path/to/your/images --rename
   
   # Organize to a custom output directory
   doggo organize /path/to/your/images --rename --output /path/to/organized
   
   # Organize in place (within the same directory)
   doggo organize /path/to/your/images --inplace

Organizing Images

Doggo can intelligently organize your images using AI-generated categories and descriptive filenames:

AI-Generated Categories

During indexing, Doggo analyzes each image and assigns it to a category (e.g., "flower", "dog", "landscape", "food"). When you run the organize command, images are automatically grouped into folders based on these categories.

Smart Renaming

With the --rename flag, Doggo generates descriptive filenames based on the image content. Instead of generic names like "IMG_001.jpg", you get meaningful names like "red_rose_garden.jpg" or "golden_retriever_park.jpg".

Organization Options

• Default: Creates an "organized" folder in the source directory
• Custom output: Use --output to specify a different destination
• In-place: Use --inplace to organize within the original directory
• Combined: Use --rename with any output option for descriptive filenames

Example Organization

Before:
📁 photos/
  ├── IMG_001.jpg (a red rose)
  ├── DSC_123.jpg (a dog in park)
  └── photo.jpg (sunset)

After:
📁 organized/
  ├── 📁 flower/
  │   └── red_rose_garden.jpg
  ├── 📁 dog/
  │   └── golden_retriever_park.jpg
  └── 📁 landscape/
      └── sunset_beach_view.jpg

How it works

• AI-Powered Indexing: Doggo scans directories for images, uses AI vision models to generate detailed descriptions and categories of each image, and converts these descriptions into vector embeddings for semantic search capabilities. Supports both OpenAI and local Ollama models.

• Vector Database Storage: The tool stores image metadata, AI-generated descriptions, categories, and vector embeddings in a local ChromaDB database, enabling fast similarity-based retrieval and intelligent organization without needing to re-process images on each search.

• Natural Language Search: Users can search for images using descriptive queries like "cute dog playing in the park" - the system converts the query to a vector embedding and finds the most semantically similar images using vector similarity search.

• Intelligent Organization: Using the AI-generated categories and descriptions, Doggo can automatically organize images into meaningful folders and generate descriptive filenames, making it easy to find and manage large collections of images.

• CLI Interface: Provides a simple command-line interface with commands for initialization (doggo init), configuration (doggo config), indexing (doggo index ), searching (doggo search "query"), and organizing (doggo organize ) with rich output formatting and progress tracking.

Configuration

Doggo supports multiple AI providers through a flexible configuration system:

Supported Providers

• OpenAI: Cloud-based AI services (default)
• Ollama: Local AI models running on your machine

Configuration Options

• --provider-url: AI provider endpoint (default: OpenAI)
• --chat-model: Model for image analysis and chat (default: gpt-4o)
• --embedding-model: Model for text embeddings (default: text-embedding-3-small)
• --api-key: API key (required for OpenAI, optional for Ollama)

Example Configurations

OpenAI with custom models:

doggo config set --provider-url "https://api.openai.com/v1" \
                 --chat-model "gpt-4" \
                 --embedding-model "text-embedding-ada-002" \
                 --api-key "sk-your-key"

Ollama with local models:

doggo config set --provider-url "http://localhost:11434/v1" \
                 --chat-model "llama3.2:3b" \
                 --embedding-model "nomic-embed-text:latest"

View current settings:

doggo config show

Provider-Specific Notes

• OpenAI: Requires API key, supports vision models for image analysis
• Ollama: No API key required, ensure Ollama is running locally
• Vision Support: Not all Ollama models support image analysis - use models like llama3.2:3b that have vision capabilities

Contributing

• Contributions are welcome! Please feel free to submit a pull request.
• See open issues for ideas.

License

MIT License - see the LICENSE file for details.