Multimodal · DSH Multimodal
yauntyour/DSH-Multimodal · productivity
DSH Multi-modular Input Plugin: Configure a separate processing model chain for different types of files (photograms/ video/ audio/ text) before the file enters the session model, using preset models to process it into Prompt Tokens (text), and then hand over to the session model.
★ 2GitHub stars
0Forks
2026-08-25Last updated
TypeScriptLanguage
MITLicense
Key features
- Settings PageAdd "Multimodal" menu page to the DSH settings and configure different file formats with wildcards
- host semi (Node)set namespace multimodal (schemastery scheme + codis. yml as base)
- Create multiple presets to configure different models for different file types (effective first match in configuration order)
- Automatically use the current default model (i.e., "compatible with its multiple mode input model") when the model chain is empty
Requirements
- Video / Audioonly text and picture content sections are currently supported in the appropriate level, while metadata briefs (file name/type/size/mime) are sent to the processing model for video/audio files, which cannot be carried directly
- If you need a transfer, use the corresponding tool in the session
Install command
dsh plugin --profile web add github:yauntyour/DSH-Multimodal