Image understanding
Argonaut790/dsh-deepseek-vision · model-adapters
Image understanding, OCR, and persistent visual evidence for text-only DeepSeek Harness models.
★ 2GitHub stars
0Forks
2026-08-24Last updated
TypeScriptLanguage
MITLicense
Key features
- DSH DeepSeek Vision is an open-source DeepSeek Harness (DSH) vision plugin that adds image understanding, full-screen OCR, and persistent visual evidence to text-only DeepSeek models without replacing the parent model.
- Selected images are sent to the configured vision provider.
- latest (default)images from the newest conversation event containing delegated images
- see_image with latest, all, and explicit image selection
- one conversation-scoped vision analyst with follow-up memory
Requirements
- an image-capable model registered in the Harness catalog
- the DSH spawn subagent provider
Install command
dsh plugin --profile web add github:Argonaut790/dsh-deepseek-vision