3ni8ma/HELIOS
Live in productionHELIOS is a browser-native, multimodal spatial operating system designed to bridge the gap between human intent and digital data. By fusing MediaPipe-powered ne
Browser-based AI OS with hand-gesture navigation and 3D spatial interface
- JavaScript51.3%
- HTML42.0%
- CSS4.9%
- Shell1.8%
1 Review
HELIOS presents an ambitious browser-based spatial workspace built from understandable modules. Rendering, hand tracking, voice recognition, audio feedback, window management, and the central visual element are separated into dedicated classes and coordinated through main.js. This makes the code easier to explore than a single large demonstration script. The project combines Three.js, MediaPipe, the Web Speech API, CSS3D windows, local session storage, and keyboard controls into a coherent prototype. Its architecture guide, gesture documentation, troubleshooting notes, feature inventory, changelog, and third-party license notices provide much more context than most projects of this size. The compatibility test page also gives users a quick way to check WebGL, camera, speech, storage, and CDN imports.
The repository layout and verification process need attention before the project can support its “production ready” language. The primary project lives inside a second HELIOS directory, and the quick-start instructions tell users to open index.html, although that file is not present. The available entry points have names such as index-standalone.html and test-init.html, which makes the correct launch path unclear. Testing is limited to browser test pages, with no package configuration, automated assertions, CI workflow, or reproducible performance measurements for claims such as 60 FPS and sub-50ms gesture latency. The compatibility page also prints a final success message even when earlier checks fail. Automated browser tests, a documented local HTTPS server command, graceful operation when camera or microphone permission is denied, and clearer privacy information would strengthen the project. The gesture callbacks should also be reviewed against the README because pinch currently provides feedback without implementing the documented window selection and dragging behavior.
