seek://map/modlens
ModLens — Vision Plugin
#vision #multimodal
TL;DR
ModLens is the first vision plugin for DeepSeek Harness: it gives the terminal agent eyes, acting as a vision bridge for every terminal. A capability the stock harness lacks, delivered as a plugin.
Proof of the plugin thesis: a whole sensory modality arrives as a community plugin, not a framework release.
Key points
- First vision plugin in the DSH ecosystem
- Designed as a vision bridge for terminal agents generally, not just DSH
- Actively maintained, with a product page at liustack.dev
FAQ
Does DSH support images natively?
The stock harness focuses on text; ModLens adds image understanding as a plugin, which is exactly the everything-is-a-plugin promise in action.
Where does it run?
As a DSH plugin; see the repo README and liustack.dev for setup details.
Is it open source?
Yes — MIT licensed.
Official references
- docsGitHub README ↗
- docsProject site ↗