Skip to content

open-agent3/agent3

Agent3

English | 中文说明

CI License: MIT

A system-level ambient AI voice agent that feels natural, fast, and close at hand.

🌐 Visit the Official Website & Download

Status: Public Alpha (v0.1.x)

📥 Installation

For average users, you don't need to touch any code. Just download the installer for your operating system:

👉 Get Agent3 for Windows / macOS / Linux

Note: After installation, you will need to provide your own API key (OpenAI or Gemini) in the tray settings to start using the voice agent.


✨ Built For Everyone

Even if you're not technical, getting started usually takes just one API key.

Agent3 is ready for early adopters, builders, and contributors who want to try a new kind of desktop voice agent. Core voice workflows already work, but the product is still evolving quickly.

📱 Tech Stack: Tauri 2.0 + Vue 3 + TypeScript + Rust + SQLite + cpal
🧠 Providers: OpenAI, Gemini (Realtime WebSockets)
🚀 Features: All-native audio processing, Rustpotter wakeword detection, transparent desktop overlay, local knowledge graph memory, and OS-level tool execution.

✅ What Works Today

  • Realtime voice conversation with OpenAI and Gemini
  • Native audio capture and playback in Rust
  • Wake word creation, activation, and detection
  • Voice switching, desktop overlay, and tray-based controls
  • Local persistence for settings, transcripts, and memory data

⚠️ Current Limits

  • This is not a polished 1.0 consumer release yet
  • Behavior may still vary across devices, microphones, and OS edge cases
  • Onboarding has improved, but some system-level flows may still need refinement
  • Expect rapid iteration and occasional breaking changes

🔒 Privacy Notes

  • Audio is captured locally by the native app
  • Settings and memory data are stored locally in SQLite
  • Realtime conversation audio and messages are sent to your configured AI provider when active
  • You should review the privacy terms of the provider you choose

⚠️ Security Warning

  • Agent3 can execute OS-level tools, including shell commands and desktop actions
  • Do not use it on machines that contain highly sensitive data unless you fully trust the setup
  • Review provider settings, prompts, and future tool permissions carefully before daily use
  • For early testing, prefer a non-critical environment or secondary machine

⚡ Prerequisites

To build and run Agent3, you need the following installed:

🛠️ Quick Start

# 1. Install dependencies
pnpm install

# 2. Run in development mode (hot-reloading enabled)
pnpm tauri dev

Note: The main window is a transparent, always-on-top click-through layer. To access configuration, look for the Agent3 icon in your system tray and open Settings.

🤝 Contributing

We welcome community contributions! Please read our CONTRIBUTING.md for details on our code of conduct, development workflow, and the process for submitting Pull Requests to us. Note: All Pull Requests should target the dev branch.

📜 License

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

About

A system-level ambient AI voice agent.

Topics

Resources

License

Code of conduct

Contributing

Stars

Watchers

Forks

Packages

 
 
 

Contributors