Contributing
Ways to help, and what a pull request needs.
Thank you for helping make Hush AI better.
Ways to help#
- Try it on your computer and report what breaks: system, microphone, meeting app and what you heard.
- Pick a roadmap item from the README and open an issue to talk about the approach first.
- Improve the docs: short, plain sentences; one idea per sentence.
Before you send a pull request#
- Read Build and release and run the tests.
- Keep audio rules: no allocation, locks or waiting in audio callbacks; blocking device calls stay on the control thread.
- Keep the regression gates. Never loosen a gate to make a change pass.
- Measure performance changes with release builds, and nothing else running.
- Do not commit recordings of anyone's voice, secrets, tokens or model weights.
- Match the style of the code around you.
cargo fmtandnpm run typecheckmust pass.
Design#
The app's look is warm white with an opal palette (peach, apricot, mint, aqua), a deep ink
for text and primary buttons, Geist type and big rounded cards. Tokens are in
app/src/styles.css. Motion is short and springy, and nothing
animates while the window is hidden.
License#
By contributing you agree that your contribution is licensed under Apache-2.0.
Something unclear or out of date? Tell us or edit this page.