deepseek-vision
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., "@deepseek-visionAnalyze the image at /home/user/photos/screenshot.png and describe what it shows."
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 Server
给 DeepSeek 等纯文本模型加"眼睛"——截图分析、图片识别,用免费的 Agnes 2.5 Flash。
功能
工具 | 说明 |
| 截屏 + 视觉分析(全屏/主屏/指定窗口) |
| 纯截屏保存 |
| 列出所有可见窗口 |
| 图片分析 |
| 列出可用模型 |
| 配置自检 |
Related MCP server: MCP Vision Server
前置条件
Node.js >= 18(下载:https://nodejs.org)
Agnes API key(获取:https://www.agnes-ai.cn,当前免费)
安装
Codex
编辑全局配置 ~/.codex/config.toml:
[mcp_servers.deepseek-vision]
command = "npx"
args = ["-y", "deepseek-vision"]
env = { VISION_API_KEY = "你的Agnes API-key" }或命令行:
codex mcp add deepseek-vision -- npx -y deepseek-vision重启 Codex。
Claude Code
编辑全局配置 ~/.claude/mcp.json:
{
"mcpServers": {
"deepseek-vision": {
"command": "npx",
"args": ["-y", "deepseek-vision"],
"env": { "VISION_API_KEY": "你的Agnes API-key" }
}
}
}或项目级 .mcp.json,内容同上。
或命令行:
claude mcp add deepseek-vision -- npx -y deepseek-vision重启 Claude Code。
opencode
编辑全局配置 ~/.config/opencode/opencode.json:
{
"mcp": {
"deepseek-vision": {
"type": "local",
"command": ["npx", "-y", "deepseek-vision"],
"enabled": true,
"environment": { "VISION_API_KEY": "你的Agnes API-key" }
}
}
}或项目级 opencode.json,内容同上。
重启 opencode。
验证
重启客户端后,对 agent 说:
调用 zen_status 检查配置返回"API 连通正常"即安装成功。
Skill(可选,让 agent 自动调用工具)
Skill 是告诉 agent 什么时候、怎么调用 MCP 工具的指令文件。没有 Skill,agent 可能不知道要调用我们的工具。
安装
把 deepseek-vision-skill/SKILL.md 复制到客户端的 skill 目录:
Claude Code:~/.claude/skills/deepseek-vision/SKILL.md
opencode:~/.config/opencode/skills/deepseek-vision/SKILL.md
# Claude Code
mkdir -p ~/.claude/skills/deepseek-vision
cp deepseek-vision-skill/SKILL.md ~/.claude/skills/deepseek-vision/
# opencode
mkdir -p ~/.config/opencode/skills/deepseek-vision
cp deepseek-vision-skill/SKILL.md ~/.config/opencode/skills/deepseek-vision/安装后重启客户端,agent 会自动发现并加载 Skill。
Skill 包含三条核心规则:
图片/截图 → 必须调用 MCP 工具
禁止用 OCR/Read 代替
根据模型 ID 判断走 MCP 还是内置多模态
使用示例
"帮我看看当前桌面在干什么"
"分析这张图片 D:\x\photo.png"
"现在有哪些窗口开着"配置项
变量 | 必填 | 默认值 | 说明 |
| ✅ | - | Agnes API key |
| 否 |
| 视觉模型 |
| 否 |
| 中国端点 |
| 否 |
| 截图保存目录 |
许可
MIT
This server cannot be installed
Maintenance
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
- Alicense-qualityBmaintenanceA drop-in MCP server that pairs long-context reasoning LLMs with vision models in description-only mode, enabling any reasoning model to 'see' images without the vision model giving advice or solutions.Last updatedMIT
- Flicense-qualityBmaintenanceA versatile MCP server that adds vision capabilities (image analysis, OCR, image/video generation) to AI models lacking native vision, with support for multiple providers and automatic task routing.Last updated
- FlicenseAqualityBmaintenanceLocal MCP server that adds multimodal capabilities to text-only models like Codex/DeepSeek, offering tools for image description, audio transcription, video analysis, image/video generation, and speech synthesis.Last updated10
- Alicense-qualityCmaintenanceMCP server that provides a 'borrowed eye' for text-only LLMs, enabling them to identify and describe local images via the Qwen VL vision model, including face recognition, scene description, OCR, and targeted visual questioning.Last updated1Apache 2.0
Related MCP Connectors
OCR, transcription, file extraction, and image generation for AI agents via MCP.
MCP server for Google Veo AI video generation
MCP server for AI dialogue using various LLM models via AceDataCloud
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/Sunlianwang/deepseek-vision'
If you have feedback or need assistance with the MCP directory API, please join our Discord server