deepseek-mcp-image
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-mcp-imageCan you describe what's in this image? /tmp/pic.jpg"
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.
deepseek-mcp-image
中文 | English
给 DeepSeek 等纯文本模型 用的识图 MCP 服务器:让不支持图片输入的 LLM 也能看图。
主模型调用 describe_image 工具并传入图片(本地路径或 http(s) URL),
服务器将图片转为 base64 data URL,交给支持图像输入的商汤多模态模型
SenseNova 6.8 Flash-Lite(sensenova-6.8-flash-lite)识别,返回文字描述。
环境要求
Node.js ≥ 18
商汤开放平台 API Key(sensenova.cn)
Related MCP server: vision-bridge-mcp
安装
git clone https://github.com/Chuyuxuan0v0/deepseek-mcp-image.git
cd deepseek-mcp-image
npm install配置环境变量
变量 | 必填 | 默认 | 说明 |
| ✅ | — | 商汤 API Key |
| — |
| API 基础地址 |
| — |
| 单张图片大小上限 |
PowerShell(Windows)示例:
$env:SENSENOVA_API_KEY = "sk-你的key"运行测试
npm test接入 MCP 客户端
标准 stdio MCP 服务器,启动命令:node <本仓库路径>/src/index.js。
Claude Desktop
claude_desktop_config.json 示例:
{
"mcpServers": {
"deepseek-mcp-image": {
"command": "node",
"args": ["<本仓库路径>/src/index.js"],
"env": { "SENSENOVA_API_KEY": "sk-你的key" }
}
}
}DeepSeek Harness(dsh)
在 profile 补丁层(如 ~/.dsh/profiles/web/cordis.patch.yml)追加一行:
- insert:
- id: mcp-vision
name: '@deepseek-ai/dsh-mcp-client'
config:
serverName: vision
transport: stdio
command: node
args: ['<本仓库路径>/src/index.js']
env:
SENSENOVA_API_KEY: !!js process.env.SENSENOVA_API_KEY ?? ''
failOnStartupError: false重启 dsh 后,主模型即可看到 mcp__vision__describe_image 工具(图片支持本地路径或 URL)。
其他客户端
Reasonix / Cursor / 通用 MCP 客户端按各自 stdio 服务器配置方式接入即可。
工具说明
describe_image
参数 | 类型 | 必填 | 默认 | 说明 |
| string | ✅ | — | 本地文件路径(绝对/相对)或 |
| string | — |
| 要问图片的问题 |
| integer | — |
| 最大生成 token 数,上限 4096 |
支持的图片格式:PNG / JPEG / GIF / WebP / BMP / SVG。
手动验收
设置好 SENSENOVA_API_KEY 后,用 SDK 客户端脚本调用一次:
cd deepseek-mcp-image && node --input-type=module -e "
import { Client } from '@modelcontextprotocol/sdk/client/index.js';
import { StdioClientTransport } from '@modelcontextprotocol/sdk/client/stdio.js';
const transport = new StdioClientTransport({
command: process.execPath,
args: ['src/index.js'],
env: { ...process.env, SENSENOVA_API_KEY: process.env.SENSENOVA_API_KEY },
});
const client = new Client({ name: 'manual', version: '0.0.1' });
await client.connect(transport);
const r = await client.callTool({ name: 'describe_image', arguments: { image: 'https://www.sensenova.cn/images/logo.png' } });
console.log(JSON.stringify(r, null, 2));
await client.close();
"预期输出:isError 为 false,content[0].text 为图片的文字描述。
限制
本地图片以 base64 data URL 内联传输,单张上限默认 20MB(
MAX_IMAGE_BYTES可调)。不缓存 API 结果;每次调用都会消耗商汤 API 额度。
finish_reason = content_filter时返回合规拦截提示(可能命中商汤内容审核)。
License
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Tools
Related MCP Servers
- FlicenseAqualityCmaintenanceEnables LLMs like DeepSeek to understand images by calling external vision models via OpenAI-compatible API. Provides tools to describe images or diagnose connectivity.2
- AlicenseAqualityBmaintenanceBridges a vision model to enable text-only models like DeepSeek to describe images, extract text, and compare images via MCP tools.512410MIT
- Flicense-qualityCmaintenanceEnables text-only language models to 'see' and describe images by calling multimodal APIs (OpenAI, Anthropic) for image analysis.
- Alicense-qualityBmaintenanceProvides a recognize_image tool that enables image recognition using DeepSeek's vision capabilities. Supports automatic login via browser or manual token configuration for server environments.175MIT
Related MCP Connectors
OCR, transcription, file extraction, and image generation for AI agents via MCP.
LLM chat, text summarization and AI image generation
Analyze images from multiple angles to extract detailed insights or quick summaries. Describe visu…
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/Chuyuxuan0v0/deepseek-mcp-image'
If you have feedback or need assistance with the MCP directory API, please join our Discord server