Show HN: Run coding agents on your own machine, steer them from anywhere (github.com)

Roamux has unveiled a new remote control platform that allows users to operate AI coding agents directly on their own machines, maintaining full control over their source code and credentials. Users can install a lightweight daemon on their computer that connects to the cloud, enabling steering and monitoring of coding agents from any authenticated device, such as a phone or browser. This setup ensures that all sensitive data, including repository files, git states, and credentials, remain securely on the user’s machine and are never transferred over the cloud. This innovation is significant for the AI/ML community as it emphasizes enhanced security and privacy while leveraging AI capabilities in coding. With Roamux, developers can initiate and manage coding tasks in real-time, approve permission requests, and track changes to their projects without the risk of exposing critical information. The architecture implements strict security measures, including localized agent runtime, authenticated routing, and a controlled command set to prevent potential vulnerabilities. By making it easier to run various agents like OpenCode, Claude Code, and Codex locally, Roamux fosters a more flexible and secure environment for AI-assisted programming, encouraging experimentation and collaboration within the community.