dsh-bundle-vision
Run the following command in DeepSeek Harness:
dsh plugin install skillre/dsh-bundle-vision
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install skillre/dsh-bundle-vision in your terminal; the full source is available at https://github.com/skillre/dsh-bundle-vision .
About this plugin
The main model in DeepSeek Harness is typically a text-only route, and when a user needs to OCR a screenshot or describe a local photo, they either forward the image to a separate model manually or patch the Harness core to bring in multimodal support. dsh-bundle-vision collapses that workflow into a single plugin install: it registers a describe_image tool that lets the text model read a local PNG, JPEG, WebP, or GIF file, commits the bytes through the built-in attachment service, fires one direct LLM request at a named multimodal route, and returns plain text to the session — the image itself never enters the main model context.
Use Cases
- OCR a local screenshot to extract text content within an image
- Ask the main model to inspect a design mockup or chart and return a description or analysis
- Switch between different multimodal routes per call without editing profile settings
Best For
- Developers using DeepSeek Harness whose main model is a text-only route
- Users who occasionally need to inspect local images without modifying core code
- Plugin users who have configured multimodal models and want tool-based access from a text workflow
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.