limccn/deepseek-vl-support

Name: limccn/deepseek-vl-support
Author: limccn

Plugin插件 Native原生 ⭐ 3 MIT Vision & Media视觉与多媒体 Models & Routing模型与路由

让DeepSeek（纯文本）模型在Claude Code和Codex中拥有**视觉**能力，通过将图像文件路由至任何兼容OpenAI的视觉端点（OpenRouter、SiliconFlow、DashScope、Ollama、llama.cpp、vLLM、LM Studio等）。零运行时依赖，MIT许可。

Project Overview项目介绍

Edited by Zhiwei Agent (zhiwei)由知微Agent(zhiwei)编辑

This is a cross-agent tool that adds image description capabilities to text-only large language models, including DeepSeek. Whenever the connected AI coding agent tries to read an image file from the local filesystem, the tool intercepts that read request, sends the image to a user-selected vision service, gets back a detailed text description of the image, and injects that description back into the content the model sees. It supports more than 20 popular AI coding agents, including DSH, Claude Code, Codex, Cursor, GitHub Copilot, VS Code, OpenCode, Trae, Qwen Code, and many others.

To install the tool, users just need to navigate to their project folder in a terminal and run one command: npx @limccn/deepseek-vl-support@latest install. The installation wizard automatically detects any supported agents on the local machine, asks just a handful of simple questions, most of which have sensible default answers that users can accept by pressing Enter. After the wizard finishes, users only need to restart their agent session for the tool to take effect, and can run a doctor command to verify the installation is working correctly. If users do not have access to a terminal, they can ask their agent to install it directly.

The tool requires Node.js version 18 or newer to run, and users need to provide an API key for their chosen vision service, which is stored only locally on the user's machine. It supports both cloud-based vision services (like Moonshot, OpenRouter, Zhipu GLM) and local options like Ollama, llama.cpp, vLLM, and LM Studio, so users can choose to run everything on their own hardware if they prefer. Descriptions of images are cached locally on disk, with a 64MB limit on the total cache size, and the whole project is released under the open source MIT license. Users can adjust settings like maximum image size via a simple CLI command any time.

本工具为纯文本大模型添加视觉理解能力，能拦截AI编码代理读取图片文件的请求，将图片发送给用户指定的视觉模型服务，获取图片的文本描述后返回给原模型，让原本无法处理图片的文本模型也能读懂截图、UI原型图、图表等内容。它支持包括DSH、Claude Code、Codex、Cursor在内的二十余种常见AI编码代理。

它面向使用纯文本大模型的开发用户，无需修改模型配置，也不需要手写配置文件，完成一次性安装配置后即可自动工作。用户只需要在项目目录执行一行安装命令，安装向导会自动检测系统中的代理，引导完成配置，全程默认选项即可完成安装，之后重启代理会话就能生效。

本工具要求Node.js 18或更高版本，用户需要准备一个视觉服务的API密钥，支持公有云服务和本地自托管选项如Ollama、llama.cpp。描述结果会缓存在本地，避免重复调用节省成本，项目采用MIT许可开源免费使用，总缓存大小限制为64MB。

View on GitHub在 GitHub 查看

Send to your DSH发给你的 DSH 安装

请帮我安装这个 DSH 插件。安装前先完成【兼容性检查 + 安全性检查】，检查通过再动手。

插件：deepseek-vl-support（limccn/deepseek-vl-support）
仓库：https://github.com/limccn/deepseek-vl-support
本站详情页：https://www.yhbd.top/plugins/limccn-deepseek-vl-support/
本站登记：类型 plugin · 归类 原生 DSH 插件 · 许可证 MIT · ⭐ 3 · 最近提交 2026-09-01 · 主语言 TypeScript

按下面顺序执行，每步先把结论告诉我，再进入下一步：

【1 兼容性检查】
① 我这边：DSH 版本、Node 版本、操作系统、当前 profile（web / desktop）。
② 读它的 README、package.json、插件 manifest，列出它要求的 DSH 版本 / Node 版本 / 操作系统 / 外部依赖 / 需要另外先装的运行时。
③ 逐条比对，结论只写「满足 / 不满足 / 未知」三种；不满足的给出可行替代方案。
④ 检查是否和我已装的插件冲突：命令名重复、skill / tool 重名、端口占用、重复注册的 MCP server。

【2 安全性检查】
① 仓库可信度：和上面「本站登记」是否一致；star / fork 数、创建时间、最近提交，是否归档或长期停更。
② 安装脚本：逐行看 package.json 的 preinstall / install / postinstall，以及 install.sh、setup.ps1 之类脚本。出现 curl|bash、下载后直接执行、混淆代码、访问与插件功能无关的域名，立刻停下来告诉我，不要继续装。
③ 依赖：列出新增依赖，标出无人维护、或与知名包拼写近似的可疑包（typosquatting）。
④ 权限与副作用：它会读写哪些目录、访问哪些域名、需要哪些 DSH 权限（filesystem / network / shell / clipboard 等），以及怎么卸载和回滚。
⑤ 如果它要求 sudo / 管理员权限，或权限明显超出功能所需，先停下来问我。

【3 安装】
上面两步没有「不满足」和「高危项」时才执行；用官方推荐方式安装，不要自行提权。

【4 汇报】
用表格输出：检查项 / 结论 / 依据 / 是否需要我决策。拿不准的一律写「未知」并说明要我怎么确认——不要猜，也不要替我决定。

Send this message to DSH in your current session: it verifies compatibility and security first (answering met / not met / unknown item by item) and only installs once everything checks out — it will stop and ask you if it finds a high-risk item. The box scrolls; the copy is the full prompt. CLI install commands may not be accurate across systems, so DSH is the safer route.把上面这条消息直接发给当前会话里的 DSH：它会先核对兼容性与安全性（逐条给「满足 / 不满足 / 未知」），确认没问题再安装，有高危项会停下来问你。框内可滚动，复制到的是完整提示词；安装命令不一定准确，发给 DSH 更稳。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意

Only 3 stars - very few users, little community feedback星标只有 3，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 @limccn/deepseek-vl-support@latest

把 limccn/deepseek-vl-support 加入你的 DSH 配置（web profile）即可启用。

READMEREADME

17.1 KB · 缓存于 2026-09-01 19:53:04 · 17.1 KB

deepseek-vl-support

中文说明 → docs/README.zh-CN.md

What this does

Some AI models (like DeepSeek) can read your files, but they cannot see pictures. Screenshots of errors, UI mockups, charts — invisible to them.

This small tool gives them "eyes". Once installed, whenever the model tries to read a picture, the tool sends it to a vision service of your choice (Moonshot, OpenRouter, SiliconFlow, Ollama …), receives a detailed text description, and hands it to the model — as if the model could see the picture.

Model reads screenshot.png
  → the tool intercepts the read
  → picture → vision service → detailed text description comes back
  → the model receives: "[Vision of screenshot.png]: <description>"
  → the model answers from the description

No model settings to change, no config files to write — it works automatically after a one-time setup. One command to install, one command to remove. MIT licensed.

Who this is for

You use a text-only model (such as DeepSeek) in any AI coding agent or IDE and want it to understand pictures: error screenshots, UI mockups, charts, photos of notes. Pick your tool in the install wizard below — there is a one-command install for every supported agent, including Claude Code, Codex, Cursor, GitHub Copilot, VS Code, OpenCode, Trae, Qwen Code, and 14 more.

Before you start (what you need)

Node.js 18 or newer — check with node -v. Not installed? Get it at https://nodejs.org.
An account at a vision service, plus its API key — a vision service is the "eyes provider": a website that looks at pictures for you. Cloud options: Moonshot, OpenRouter, MiniMax, Zhipu GLM, StepFun, OpenCode Zen, SiliconFlow, DashScope. Free local options (run on your own computer): Ollama, llama.cpp, vLLM, LM Studio. The API key is a secret code from that service (usually under "API keys"); the installer asks for it once and stores it only on your computer.
Your AI agent installed — any of the supported ones below.

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 →

渲染基于仓库 limccn/deepseek-vl-support 的 README.md。在 GitHub 查看原文 →