dsh-live-voice
Safety-first DeepSeek Harness live voice preview with exact-session consent and one bounded manual audio turn
安装
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:Jstn-1g/dsh-live-voice说明文档
阅读完整 README ↗Development configuration
The bundle inserts the compatibility-stable guarded-live-voice Cordis row
with the dsh-live-voice package name. A profile override will
need an Alibaba Cloud Model Studio workspace id before provider authorization:
- id: guarded-live-voice
name: dsh-live-voice
config:
credentialRef: DASHSCOPE_API_KEY
dashscopeWorkspaceId: your-workspace-id
model: qwen-audio-3.0-realtime-plus
route: /guarded-voice
trustedHosts: localhost,127.0.0.1,[::1]
maxConnections: 8
After accepted disclosure, the preview opens the allowlisted Qwen endpoint for one exact-session manual turn and resolves the credential only on the Host. A second explicit button starts browser capture; finishing it commits only that bounded provider turn. This path is fake-tested but still needs a packed Desktop smoke and credential-backed Qwen roundtrip.
Configure the credential through DSH's credential provider. Never put a secret
value in cordis.patch.yml, browser storage, an issue, or a log.