Context-aware dictation for DeepSeek Harness
franksong2702/dsh-dictate · skills
The official DSH 0.1.2-alpha.2 package is compatible with the current client extension contract and the minimum compatible version is raised to 0.1.2-alpha.2.
★ 2GitHub stars
0Forks
2026-09-07Last updated
TypeScriptLanguage
MITLicense
Key features
- Composer IntegratorThe microphone is located directly in the Composer Toolbar
- Depth integration with context-based voice input for DeepSeek Harness Composer.
- You can separately enable the " Hold down the mouse recording" in settings: Press the mouse 500 ms to start the recording, release it and end with the final result written into the input box. It will not be sent automatically.
- Compatible baseline and lock-in depend on the official DSH 0.1.2-alpha.3
- Web Speech's real-time text or local endpoint recording and transliteration status is shown above Composer rather than in the standalone suspended component.
Requirements
- The minimum compatible version of 4.0-alpha.8 is DSH 0.1.2-alpha.3
- The default mode for Web Speech requires a Web or Edge Web Page API.
Install command
dsh plugin --profile web add github:franksong2702/dsh-dictate