| Version | Supported |
|---|---|
| 0.1.x | Yes |
Fixes go into the latest 0.1.x release.
Fourier Samples is a command-line tool that runs on your own machine. It runs no server, opens
no listening port and sends no telemetry. It reads your sample library and writes only its own
folders (its home, ~/.fourier by default, with its logs and report pages; its config,
~/.config/fourier; the master, renders and releases), the CLAP model in the Hugging Face cache
(~/.cache/huggingface), the device card you name to fourier sync, and the copies fourier tools import-folder adds to a library folder. Its network use: fourier setup installing
PyTorch and Transformers (from PyPI, or PyTorch's own index for a CPU-only build) when you say
yes, and offering uv self update when uv is too old for that; downloading the CLAP model from
Hugging Face once, pinned to one revision; and, when you turn folder descriptions on, a local
Ollama server on localhost. The report page (fourier open report) is a local file that plays
the master's own files; it loads nothing from the network.
Issues worth reporting privately include a way for a crafted audio file, config file, device profile or plug-in setting to run code, read or write outside the folders above, or delete or change files in a library; and a path a build, render, publish or sync would write that escapes its target folder.
Please report it privately through GitHub: on the repository's Security tab, choose
Report a vulnerability (a private security advisory). Don't open a public issue for it.
Include the version (fourier --version), your OS and Python version, and the steps or files
that show the problem.
You can expect an acknowledgment within a week. We'll work on a fix in the private advisory, credit you in the release notes unless you'd rather not be named, and publish the advisory when a fixed release is out.