linenxi-ctrl/dsh-vision 预览 preview

linenxi-ctrl/dsh-vision

为 DeepSeek Harness 增加外挂识图模型:圆形鲸鱼按钮、发送图片识图自动回传、模型自主截图+识图工具、多协议自动适配、小白一键安装(未装 Node.js 自动下载)

Project Overview项目介绍

dsh-vision is a native plugin built exclusively for DeepSeek Harness (DSH) that adds external vision model support to DSH. It enables non-vision-capable large language models running on DSH to understand images and screenshots by routing vision requests to a user-configured external vision model. It supports both user-initiated manual image uploads for recognition and agent-initiated automated screenshot capture and recognition. Users can configure all connection settings directly from the DSH web UI via a floating drag-to-move settings panel added by the plugin. It natively adapts to four common vision API protocols and supports custom templates for niche endpoints, so users can connect almost any vision API.

The typical workflow for manual recognition starts with opening a chat session in DSH, then clicking the whale-shaped floating button on the bottom right of the page to open the settings panel. Users select the image they want recognized, then the plugin converts it to base64 and sends the recognition request to the external API through the DSH host, avoiding CORS errors common in browser-side requests. Once the recognition is complete, the plugin automatically injects the resulting text into the current chat session, so the LLM can respond based on the image content without any manual copying or pasting from the user.

There are two supported installation methods: npm installation (recommended for standard setups) and manual offline installation for users without pnpm or who prefer offline setups. On Windows, double-clicking the provided install batch file will handle the entire setup, and it automatically downloads a portable Node.js build from a Chinese mirror if Node.js is not already installed. On macOS and Linux, the provided install shell script handles the entire setup automatically. All configuration is done automatically, no manual edits to config files are required, and one-click uninstall scripts are provided for all platforms. The plugin is released under the open-source MIT license.

这是专门为 DeepSeek Harness 开发的原生插件,作用是为原本不支持视觉能力的大语言模型添加外挂识图能力,支持用户手动上传图片识别,也支持代理自主调用工具截图识别,可通过自定义配置的外部视觉模型完成内容解析。插件内置了 OpenAI、Anthropic、Google Gemini 等主流接口的协议适配,能自动探测协议类型,还支持自定义模板适配各类小众接口。

用户可通过网页右下角可拖动的鲸鱼配置按钮打开面板,设置外接识图模型的 API 地址、密钥、模型名、提示词、代理和超时参数,点击发送图片后,插件会自动调用外部模型识别,再将识别结果自动注入当前对话,无需用户手动复制粘贴。适合需要给非视觉 DeepSeek 模型添加视觉能力的 DSH 用户使用。

插件支持 npm 标准安装和手动离线安装两种方式,Windows 系统可双击安装脚本,未安装 Node.js 时脚本会自动下载免安装版,macOS/Linux 可通过 shell 脚本一键完成配置,安装全程无需手动修改配置文件,项目采用 MIT 开源许可,可免费使用和修改。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 9 stars - very few users, little community feedback星标只有 9,几乎没人在用,遇到问题缺少社区反馈
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时(Electron / Python / Docker 等)是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突:命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致,是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需(filesystem / network / shell / clipboard)
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限,以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照,不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装(适合开发者)

CLI Install命令行安装

dsh plugin --profile web add @linenxi-ctrl/dsh-vision

把 linenxi-ctrl/dsh-vision 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

dsh-vision —— DeepSeek Harness 外挂识图模型插件

License Platform DSH Version npm

为 DeepSeek Harness 增加「外挂识图模型」能力:让本来不具备视觉能力的模型,通过一个可自定义地址/密钥/提示词的外部视觉模型来「看懂」图片与屏幕。

功能

  1. 网页配置按钮与面板:页面右下角出现一个DeepSeek 鲸鱼圆形按钮(可拖动),点击即可配置外挂识图模型的 API 地址、密钥、模型名、识图提示词(skill)、代理与超时。
  2. 发送图片识图并自动回传:点鲸鱼按钮打开面板,点「📤 发送图片」选图,插件会先把它发给外挂识图模型,等识别完成后把识别文本自动作为消息发回当前会话(无需手动复制粘贴),DeepSeek 基于识别文本作答。
  3. 模型自己截图 + 识图:插件为 agent 注入 screenshot(截屏)与 recognize_image(识图)两个工具,并注入提示词,模型可自行「截图 → 识图 → 等待结果」。
  4. 自动适配识图 API 协议:内置 OpenAI Chat Completions、OpenAI Responses、Anthropic Messages、Google Gemini 四种协议,并按 apiBase 自动探测;另有 custom 模板协议适配任意长尾接口。

文件结构

dsh-vision/
├── install.bat        # Windows 一键安装(双击)
├── install.sh         # macOS/Linux 一键安装
├── install.mjs        # 安装脚本本体(npm 场景只做 agent 工具平面;目录场景全自动)
├── bootstrap-node.ps1 # Windows 引导脚本:未装 Node.js 时从国内镜像自动下载免安装版
├── package.json       # 包定义(dsh.bundle + dsh.client 声明;tool 为独立子路径)
├── cordis.patch.yml   # 插件挂载声明(dsh.bundle.patch 自动应用到 profile layer)
├── lib/
│   ├── index.js       # host 平面插件:识图服务 + 协议适配 + settings 配置 + HTTP 路由
│   ├── tool.js        # agent 工具插件:recognize_image / screenshot + 提示词注入
│   └── client.js      # 客户端插件:鲸鱼按钮 / 配置面板 / 发送图片识图 / 自动回传
└── README.md

工作原理

[用户点鲸鱼按钮选图]                [模型调用工具]
      │                               │
      ▼                               ▼
  client 转 base64 发送          screenshot 工具截屏
      │                               │
      ▼                               ▼
  POST /api/vision/recognize    recognize_image 工具
      │                               │
      ▼                               ▼
  host 插件 ctx.vision 服务 ──► 协议自动适配后调用外挂识图 API
      │                               │
      ▼                               ▼
  识别文本 → 自动注入当前会话    识别文本返回给模型

识图请求在 host(Node)侧发起,因此不受浏览器 CORS 限制;图片请求走同源 /api/vision/recognize,同样无 CORS 问题。

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要,完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

← 上一个 Prev dsh-archify 下一个 Next dshscan →