dsh-vision-fix-lmstudio
Run the following command in DeepSeek Harness:
dsh plugin install M0R1C/dsh-vision-fix-lmstudio
Paste the following prompt into your AI chat to install this plugin:
Install it in DeepSeek Harness by running dsh plugin install M0R1C/dsh-vision-fix-lmstudio; the full source is at https://github.com/M0R1C/dsh-vision-fix-lmstudio .
About this plugin
Running both local vision models in LM Studio and cloud APIs from a single DeepSeek Harness instance makes image format mismatches the most frequent cause of 400 errors: local providers expect JPEG while cloud services prefer WebP or PNG, and juggling between them is both tedious and error-prone.
dsh-vision-fix-lmstudio intercepts outgoing requests and converts images to the format each model expects using the sharp library, right before they reach the provider. You simply tick the models that need conversion in the DSH settings panel, with per-provider bulk toggles and wildcard pattern support, then send your message as usual. No proxy server, no manual re-encoding, no per-request workaround.
Built for anyone who mixes local and cloud vision models in DeepSeek Harness and wants a one-time setup that permanently eliminates the format compatibility headache.
Use Cases
- Sending WebP images to LM Studio vision models auto-converts them to JPEG, eliminating the 400 encoding error.
- Mixing local and cloud vision providers in DeepSeek Harness without manually switching image formats per provider.
- Sending image-bearing messages without re-encoding files by hand or running a proxy service first.
Best For
- Users running local vision models via LM Studio inside DeepSeek Harness.
- Developers mixing local and cloud vision providers who hit format errors frequently.
- Heavy vision-model users who want a one-time setup to permanently end image format headaches.
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.