Okonomi

Okonomi is an Agent meant for experimenting with creative image creation.

Features

• Creative Strategy Tools: Leverages multiple creative thinking techniques to generate unique prompts
• Multi-modal Image Generation: Support for text-to-image, image-to-image, image editing, and style transfer
• LoRA Integration: Dynamic loading and combination of LoRA models for enhanced control
• Memory System: ChromaDB-based memory for learning from past generations
• Automatic Evaluation: Built-in evaluation system for prompt and image quality
• MCP Tools: Modular tool architecture for extensibility

Installation

Prerequisites

• Python 3.11 or higher
• uv package manager (recommended) or pip

Clone the repository

git clone https://github.com/dn6/okonomi.git
cd okonomi
uv pip install -e .

Configuration

Required Environment Variables

Create a .env file in your project root with the following variables or set them manually in your environment.

# Hugging Face API Token (Required to use LLMs)
# Get from: https://huggingface.co/settings/tokens
HF_TOKEN=your_huggingface_token

# Fal.ai API Key (Required for Image Generation)
# Get from: https://fal.ai/dashboard/keys
FAL_KEY=your_fal_api_key

# ChromaDB API Key (Optional)
# Only required for cloud-hosted ChromaDB. Leave unset for local storage
CHROMA_API_KEY=your_chroma_api_key

Optional Environment Variables

# Enable observability/tracing (Optional)
LOG_CREATE_AGENT_TRACES=false

# Custom configuration file path (Optional)
OKONOMI_CONFIG=/path/to/custom/config.json

Configuring Memory

Generated images are available through FAL's CDN. You can also save the generated images to a Hugging Face dataset by creating a dataset repository and setting the image_repo_id to your repo id in the config.json file.

Usage

Interactive Mode

# Start interactive session
okonomi

# You'll be prompted to enter your image generation request

Direct Command

# Generate an image with a direct command
okonomi "Create a cyberpunk city at sunset"

# Or using the --task flag
okonomi --task "Design a futuristic logo"

Command-Line Options

# Disable automatic evaluation (enable manual review of generated images)
okonomi --eval "Generate a landscape painting"

# Disable automatic planning (enable plan approval prompts)
okonomi --plan "Create a character portrait"

# Use custom configuration file
okonomi --config custom-config.json "Generate artwork"

Configuration File

The default configuration is in config.json. Key settings include:

• Agent Settings: Model selection, temperature, planning interval
• Inference Providers: HuggingFace and Fal.ai model configurations
• Memory Settings: ChromaDB collections for storing evaluations and LoRAs
• Tool Settings: MCP tool configurations and callbacks

Custom Configuration

You can create a custom configuration file and specify it:

okonomi --config my-config.json "Generate image"

Or set a default via environment variable:

export OKONOMI_CONFIG=/path/to/my-config.json
okonomi "Generate image"

MCP Tools

Okonomi includes several MCP (Model Context Protocol) tools:

• prompt.py: Creative prompt generation and image description
• image.py: Image generation tools (text-to-image, image-to-image, editing)
• memory.py: Memory management and LoRA discovery
• strategies.py: Creative strategy selection

Troubleshooting

Examples

Basic Text-to-Image

okonomi "A serene mountain landscape at dawn with mist"

With Creative Strategy

okonomi "Transform the concept of time into a visual metaphor"

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Support

For issues, questions, or suggestions, please open an issue on GitHub.