Check a plugin
HomeVision & Multimodal

Vision & Multimodal plugins

DSH vision plugins add image understanding, OCR, visual grounding, screenshot analysis, and multimodal workflows to DeepSeek Harness. They are how a text-only model reads an interface, a diagram, or a screenshot.

3 reviewed · 189 cataloged

Recommended projects

Cataloged projects

Repositories with traceable evidence of a DeepSeek Harness relationship. They have not been reviewed, install tested, or security checked — the signal below each one is the whole of what the registry currently knows.

How cataloging works
Cataloged

Unlimited-OCR for DeepSeek Harness with a native tool and GUI configuration

README documents a DSH install command Bundle manifest found in repository Package metadata declares a DSH dependency
1TypeScriptApache-2.0Run a check
Cataloged

dsh 中可以使用插件来调用gpt-image2生成图片

README documents a DSH install command Bundle manifest found in repository Package metadata declares a DSH dependency
1PythonMITRun a check
Cataloged

Visual plugin for dsh /rollback: trajectory anchor badges with click-to-rollback. Data layer ready; native-node rendering planned.

README documents a DSH install command Bundle manifest found in repository
1TypeScriptMITRun a check

meow-file-view

meimiaoji-creator
Cataloged

meow-file-view 解决「AI 改完文件，我却要切到 VS Code 才能看它到底改了什么」的问题。它把一个轻量文件查看器直接嵌进 DSH 对话窗：左侧目录树懒加载浏览工作区，右侧预览 Markdown（含 TOC / mermaid / 图片）、源码高亮、甚至直接编辑保存；模型每回合产出/修改的文件会以 chips 行挂在回合尾部，点一下直达该文件，命中 git 变更还能一键看 diff。 零 DSH 源码改动——独立 bundle，经 dsh 插件 --profile web add 装进 web profile，与官方插件平级共存。

README documents a DSH install command Bundle manifest found in repository
1JavaScriptMITRun a check
Cataloged

DeepSeek harness多模态插件，接近原生体验。

README documents a DSH install command
1JavaScriptMITRun a check

dsh-background

liuliyisui
Cataloged

DeepSeek Harness Web GUI 的自定义背景插件：图片 / 动图 / 视频 / 内置极光渐变，磨砂玻璃质感，配浮动控制面板

Bundle manifest found in repository Package metadata declares a DSH dependency
1JavaScriptMITRun a check

dsh-pdf

henryxiao709
Cataloged

DSH-PDF插件，让 AI 助手读取任意大小的 PDF 文件： 通过 pdfjs-dist 提取完整 Unicode 文本层（中文、英文及其它文字系统），并对扫描件/图片页自动 OCR， 手写笔记也能变成可读文本。DSH-PDF plugin — read any-size PDFs in DeepSeek Harness: full Unicode text (Chinese/English) via pdfjs-dist + automatic OCR (Windows WinRT / tesseract.js) for scanned pages. MIT.

Bundle manifest found in repository Package metadata declares a DSH dependency
1JavaScriptMITRun a check
Cataloged

DSH plugin: vision_read — route image reading to a dedicated vision model (e.g. Kimi K3) so text-only agents can see images

README documents a DSH install command Bundle manifest found in repository Package metadata declares a DSH dependency
1TypeScriptMITRun a check
Cataloged

修改DSH的背景，支持静态动态背景，支持网页图片视频，支持修改透明度

README documents a DSH install command Bundle manifest found in repository Package metadata declares a DSH dependency
1TypeScriptMITRun a check

Vision & Multimodal plugins: common questions

What is a DSH vision plugin?
A plugin that turns images into something the model can reason about — extracted text, layout structure, or a described scene.
Can DeepSeek Harness process images without a plugin?
Only when the selected model is itself multimodal. A vision plugin is what lets a text-only model work from images.
How do vision plugins handle my images?
It varies. Some run entirely locally, others send images to a hosted service. Each record lists the declared network permissions.