dsh-deepseek-vision-router
Run the following command in DeepSeek Harness:
dsh plugin install mochgolf/dsh-deepseek-vision-router
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install mochgolf/dsh-deepseek-vision-router inside DeepSeek Harness to install this plugin; see https://github.com/mochgolf/dsh-deepseek-vision-router for details
About this plugin
The DeepSeek main agent is a text-only model and cannot natively understand screenshots, charts, or photos pasted into the DSH chat composer. dsh-deepseek-vision-router adds a separate vision route to DeepSeek Harness, letting the text agent accept images and continue reasoning without swapping out the primary model.
Image bytes are forwarded to a vision provider configured in DSH's Models settings; the resulting text description is handed back to DeepSeek for normal conversation flow. The official DeepSeek adapter still owns inference, tool calls, streaming, and retries. Descriptions are marked as untrusted content before reaching the model so text embedded in an image is not promoted to system instructions. Results enjoy a bounded, process-local cache to avoid redundant external calls.
Ideal for daily users of the DSH DeepSeek text agent who occasionally need image understanding. Simply add a vision-capable provider in Models settings, pick the vision model in the plugin panel, and start pasting images into the chat composer. No code required.
Use Cases
- Paste screenshots or charts into the DSH chat composer and let DeepSeek answer questions about image content
- Process tickets or emails containing photos without switching to another multimodal model
- Drop a code screenshot into the conversation and let the agent suggest fixes with full context
Best For
- Developers using the DSH DeepSeek text agent daily
- Ops engineers who occasionally need image analysis without switching primary models
- Efficiency-focused users who want unified text and image handling inside DSH
Related Plugins
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.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.