dsh-qwen-voice
Run the following command in DeepSeek Harness:
dsh plugin install leaveimagination/dsh-qwen-voice
Paste the following prompt into your AI chat to install this plugin:
Install the plugin in DeepSeek Harness by running the dsh plugin install command; the source repository is https://github.com/leaveimagination/dsh-qwen-voice
About this plugin
Running multiple Agent sessions in parallel inside DeepSeek Harness Web usually means bouncing between tabs and retyping prompts, constantly breaking your flow. dsh-qwen-voice puts a persistent voice orb in the page corner and shifts interaction from typing to speaking: create, continue, and dispatch DSH sessions with natural language, all within a single browser tab, with the realtime audio link surviving every session switch.
Powered by the Qwen Audio Agent realtime engine and bridged through ACP, the plugin maps spoken intent to Harness session operations. Issue several instructions to one Coordinator session and tasks fan out to parallel sub-sessions; the floating panel reports each task status as it completes, supports mid-flight interruption with confirmed Gateway results, uses playback-state acks to avoid redundant announcements, and announces each finished task immediately rather than waiting for the whole batch.
Built for developers driving multiple workstreams in DeepSeek Harness Web who prefer voice over manual input. Requires a locally running Harness Web instance and a DashScope API key for the realtime voice front-end.
Use Cases
- Create and dispatch multiple Agent tasks with a single voice command without switching tabs
- Interrupt and redirect in-flight tasks on the fly during a live voice session
- Monitor per-session progress and status at a glance from the floating panel
Best For
- Developers running multiple Agent workflows in parallel within DeepSeek Harness Web
- Operators who prefer voice-first interaction over manual typing
- Engineers who want natural-language commands mapped directly to multi-session orchestration
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.