vision-mcp
Allows using local Ollama vision models (e.g., qwen2.5vl-vision) for image recognition and UI grounding analysis, enabling offline and privacy-preserving processing.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@vision-mcpWhat text and objects are in this image? /Users/me/photo.png"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
vision-mcp
给纯文本模型(如 DeepSeek)补上"看图"能力的 MCP 服务器 | Vision MCP server: image recognition for text-only models (DeepSeek, etc.)
这是什么 / What
vision-mcp 是一个标准的 MCP 服务器:通过 MCP 工具为不支持图片输入的大模型(如 deepseek-v4-flash)提供图像识别能力。
不依赖任何特定客户端——Codex、Claude Code、Cursor、Cline、opencode 等任何支持 MCP 的 agent 都能用,别人不需要安装 opencode。
默认后端:OpenCode Zen MiMo V2.5 Free(
mimo-v2.5-free)——免费、无需 API key,识别质量好。可选后端:本地 Ollama(
qwen2.5vl-vision)——离线/隐私场景,图片不出本机。可选:任何 Zen 模型列表里的模型(如
gpt-5.6-luna、qwen3.6-plus),直接传裸模型名即可。
Related MCP server: videre-mcp
工具 / Tools
工具 | 说明 |
| 识别/描述图片(内容、文字、颜色、布局、物体位置)。输入支持:本地路径、http(s) URL、 |
| 用本地 Ollama 分析 UI 截图,输出交互元素及归一化坐标(0–1000,多轮投票合并)。用于 GUI 自动化。 |
| 健康检查:Zen 可达性 + Ollama 可达性,列出可用模型。 |
安装运行 / Install & run
npm install -g vision-mcp # 或本地:npm run build && node dist/index.js
vision-mcp # 启动 stdio MCP 服务器
vision-mcp doctor # 诊断报告(后端可达性/模型列表)或者不安装直接用:
npx -y vision-mcp要求 Node.js 18+。
客户端配置 / Client configuration
Codex(~/.codex/config.toml):
[mcp_servers.vision]
command = "npx"
args = ["-y", "vision-mcp"]Claude Code:
claude mcp add vision -- npx -y vision-mcpopencode(opencode.json):
{
"mcp": {
"vision": {
"type": "local",
"command": ["npx", "-y", "vision-mcp"],
"enabled": true
}
}
}Cursor:设置 → MCP → 添加 → 命令:npx -y vision-mcp
使用 / Usage
直接让你的 agent "看一下 / 描述 / 读一下这张图",把路径或 URL 给它:
描述 C:\path\to\image.png —— 图里有什么文字和物体?agent 会自动调用 vision_recognize。GUI 自动化场景:vision_grounding —— "列出 screenshot.png 里的按钮及其位置"。
配置(环境变量)/ Configuration
变量 | 默认值 | 说明 |
|
| OpenCode Zen 端点 |
|
| 默认 Zen 模型(免费) |
| — | 可选,使用付费 Zen 模型时需要 |
|
| 本地 Ollama 端点 |
|
| 本地 Ollama 视觉模型 |
|
| 单次请求超时 |
|
| grounding 投票轮数 |
vision_recognize 传 model: "local" 强制走本地 Ollama;否则默认用 Zen 模型。任意裸模型名(如 gpt-5.6-luna)会直接透传给 Zen。
隐私说明 / Privacy
默认免费模型 mimo-v2.5-free 为尽力提供,可能随时变更;免费期间数据可能用于改进模型。敏感图片请配置 API key 或使用 model: "local" 全本地处理。
开发 / Development
npm install
npm run build # tsc -> dist/
node scripts/smoke-test.mjs # 端到端 MCP 客户端冒烟测试(需联网,可选 Ollama)
npm run doctor # 同 node dist/index.js doctor许可证 / License
MIT
This server cannot be installed
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- AlicenseAqualityBmaintenanceBridges a vision model to enable text-only models like DeepSeek to describe images, extract text, and compare images via MCP tools.513410MIT
- AlicenseAqualityBmaintenanceBridges vision models to text-only coding models using Florence-2, enabling non-vision LLMs to describe images, extract text, and analyze screenshots via MCP tools.6MIT
- Alicense-qualityBmaintenanceMulti-backend AI vision for MCP agents. Analyze images, screenshots, and documents using local Ollama models or cloud APIs like OpenAI, Google Gemini, and OpenRouter.MIT
- Alicense-qualityAmaintenanceLocal vision-capable MCP server that lets AI agents describe screenshots, UI, charts, and photos via vision and OCR tools, with support for multiple providers and automatic fallback.3MIT
Related MCP Connectors
OCR, transcription, file extraction, and image generation for AI agents via MCP.
User-owned memory for AI agents, Copilot, Claude, IDEs, CLIs, and chat apps over remote MCP.
Screenshot, diff, audit and sitemap-capture any web page — 5 MCP tools for AI agents.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/miaomiaozii/vision-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server