Installation¶
Savanty can be installed in several ways depending on your needs.
Requirements¶
- Python 3.10, 3.11, 3.12, or 3.13
- An OpenAI API key for LLM access
Installation Methods¶
Use uvx to run Savanty directly without permanent installation:
# Run the desktop app
uvx --from 'savanty[desktop]' savanty-desktop
# Run the web interface
uvx savanty --web
# Solve from command line
uvx savanty -p "Your problem description"
To install as a persistent tool:
Install with pip for permanent installation:
Configuration¶
Required: OpenAI API Key¶
Savanty requires an OpenAI API key to function:
Add this to your shell profile (~/.bashrc, ~/.zshrc, etc.) for persistence.
Optional Settings¶
| Variable | Default | Description |
|---|---|---|
SAVANTY_LLM_MODEL |
openai/gpt-4o |
LLM model to use |
SAVANTY_PORT |
8000 |
Web server port |
SAVANTY_LOG_LEVEL |
INFO |
Logging verbosity |
See Configuration Reference for all options.
Verifying Installation¶
Check that Savanty is installed correctly:
# Check version
savanty --version
# Check help
savanty --help
# Test with a simple problem
savanty -p "Assign task A to worker 1 or worker 2"
Platform Notes¶
Linux¶
Savanty works on most Linux distributions. Ensure you have Python 3.10+ installed:
macOS¶
Works on macOS 11 (Big Sur) and later. Install Python via Homebrew if needed:
Windows¶
Works on Windows 10 and later. Install Python from python.org.
For the desktop app, ensure you have the Visual C++ Redistributable installed.
Troubleshooting Installation¶
"No module named 'savanty'"¶
The package isn't installed in your current Python environment:
"OPENAI_API_KEY not set"¶
Export your API key before running:
Desktop app won't start¶
The Slint UI library may need additional setup:
Permission errors on Linux¶
You may need to adjust permissions or use a virtual environment:
Next Steps¶
- Getting Started - Solve your first problem
- User Guide - Learn all interface options
- Configuration - All configuration options