dsh-voice

Terminal agents often fall short in two everyday scenarios: dictating instructions when away from the keyboard and reading lengthy outputs during tasks. dsh-voice addresses these pain points by adding voice interaction capabilities to DeepSeek Harness, making the terminal experience more intuitive and efficient.

At its core, dsh-voice provides two key tools: transcribe, which converts dictated audio into user messages, and speak, which reads agent replies aloud. An added convenience is the /voice command, allowing you to toggle auto-narration on the fly. The plugin is designed with a local-first approach—audio files are stored locally by default, and no data leaves your machine unless you explicitly configure a cloud backend, ensuring strong privacy and security.

This plugin is ideal for terminal users who need hands-free interaction. Developers juggling multiple tasks or those looking to streamline communication with their agents will find its voice features a significant productivity boost. It's also perfect for privacy-conscious users, thanks to its offline processing capabilities. In essence, dsh-voice is a lightweight yet powerful enhancement that makes terminal agent interactions more natural and accessible.