┌───┐   ██████   █████   ██████   ██  ██
│ • │   █    █   █   █     ██     ██████
│ • │   ██████   █████     ██       ██
└───┘   █        █   █     ██       ██

PATY is entirely local. And therefore, is entirely free.

Install & run

Prerequisites: uv and portaudio. The happy path is for MacOS on Apple Silicon. CUDA coming soon.

curl -LsSf https://astral.sh/uv/install.sh | sh   # if you don't already have uv
brew install portaudio
uv tool install paty
paty

Bare paty (no subcommand) shows a DOS-style boot screen while the agent warms up, then auto-hands off to the live TUI once it's ready. Use paty run if you want the raw agent without the boot UI / TUI. Either form with no config argument loads a bundled default config; the first run will detect your platform and tell you which extra to install for local inference:

uv tool install 'paty[mlx]'    # Apple Silicon
uv tool install 'paty[cuda]'   # NVIDIA GPU -- coming soon

Then paty again. First launch downloads the LLM and STT models (a few GB) and is slow; subsequent runs reuse the Hugging Face cache.

CUDA users also need a Kokoro FastAPI server on localhost:8880 for TTS — Apple Silicon runs Kokoro in-process.

See cli/README.md for the config schema, CLI commands, hardware profiles, the event bus, and dev setup.

Building locally

To run paty from a clone of this repo (instead of uv tool install paty):

git clone https://github.com/PATYai/PATY.git
cd PATY/cli
uv sync --extra mlx            # Apple Silicon
# uv sync --extra cuda         # NVIDIA GPU -- coming soon
uv run paty                    # boot screen → TUI; use `uv run paty run` for the raw agent

Notes:

• The base uv sync does not install an inference backend — pick --extra mlx (or --extra cuda once CUDA lands). Without one, paty run exits with "No backend installed."
• Add --extra dev if you also want to run tests/lint (pytest, ruff).
• cuda builds llama-cpp-python from source and needs a working C/C++ + CUDA toolchain.
• CUDA users also need a Kokoro FastAPI server on localhost:8880 for TTS; Apple Silicon runs Kokoro in-process.
• First run downloads the LLM + STT weights (a few GB) into the Hugging Face cache.

PAKs (Personality Augmentation Kits)

Don't like the voice? Run paty pak switch. nova is bundled.

paty pak switch nova

Themes

PATY ships with a TUI. Run it in a separate window to mute or chat.

paty bus tui

License

This project is licensed under the MIT License - see the LICENSE file for details.