Local-first full-duplex voice for DeepSeek Harness
PiyotaHu/muxiva-dsh-voice · skills
Local-first full-duplex voice for DeepSeek Harness, orchestrated by Muxiva.
★ 2GitHub stars
0Forks
2026-08-16Last updated
PythonLanguage
Apache-2.0License
Key features
- Muxiva owns the real-time pipeline: typed Frames, bounded queues, backpressure, Signals, cancellation and observability.
- The speech bridge binds only to loopback and accepts one active client.
- DSH owns the Agentsession history, model selection, tools, permissions and Web UI.
- The plugin owns only the bridgeadditive DSH UI, a versioned loopback protocol and project Node Packs.
- Neither upstream repository is patched.
Install command
dsh plugin --profile web add github:PiyotaHu/muxiva-dsh-voice