AuroraSpeak is a lightweight web client that lets you chat with locally configured AI services through a clean, responsive interface. The focus is on clarity, speed, and full transparency over which model is responding.
- Seamless single-page UI with instant model switching.
- Supports both conversational and image-style prompts.
- Friendly error messaging with built-in retry guidance.
- Works entirely in the browser — no server setup required.
- Clone or download the project files.
- Open
index.htmlin your browser. - Configure your preferred models inside
models.json. - Start chatting instantly.
index.html— main application shell.app.js— chat orchestration, model handling, and rendering logic.style.css— layout, theming, and responsive design rules.models.json— curated list of model identifiers shown in the UI.new-feature.html— experimental playground for upcoming UI ideas.
- Add or remove model identifiers by editing
models.json. - Toggle default model selection through the
defaultModelfield inapp.js. - Adjust theme colors and spacing via CSS variables in
style.css.
- Fork the project or open an issue describing the change you have in mind.
- Keep pull requests focused on a single improvement.
- Include a short summary of manual testing or screenshots where appropriate.
- Add optional voice input for hands-free sessions.
- Expand support for streamed responses.
- Introduce offline-first caching for frequently used prompts.
This project is released under the terms of the included LICENSE.