Add vision to text-only DeepSeek via a view_image tool that bridges any OpenAI-compatible VLM.
dsh-vision is a DeepSeek Harness ecosystem companion app maintained by william-jin-cmu. Add vision to text-only DeepSeek via a view_image tool that bridges any OpenAI-compatible VLM.
Source code and usage instructions are available at https://github.com/william-jin-cmu/dsh-vision. Follow the repository README for the correct setup steps.
dsh-vision is a community open-source project released under the BSD-3-Clause license. Review the repository license and documentation before use.
UI Enhancements
A Claude Code–style full-screen TUI plugin for DeepSeek Harness: pixel-whale top bar, live working-status line, streaming thought expansion, double-Esc time rollback, and a context progress bar with a TPS gauge. One-command npm install; mounts as a pure plugin with zero core changes.
View details → 
ccch1mneyyy
UI Enhancements
This build declares these DSH releases, each verified on an isolated instance (its own DSH HOME, dependencies pinned to that release, the full check suite):
View details → 
6mikao9
UI Enhancements
Open-source, community-driven agent harness
View details → 
Hmbown