vision-mcp
Provides vision capabilities to text-only LLMs via MCP, enabling image understanding, Q&A, OCR, and image processing through cloud multimodal APIs.
README
Vision MCP
让原本不具备视觉能力的纯文本模型,通过 MCP(Model Context Protocol)调用云端多模态模型,获得图像理解、图像问答、OCR、图像处理等视觉能力。
纯文本 LLM 无法直接"看"图,但可以通过 MCP 把这个 server 暴露成工具:
文本模型 ──MCP调用──▶ vision-mcp ──▶ OpenAI / Gemini / Claude(多模态 API)
│
├──▶ 图像理解(描述、问答、OCR)
└──▶ 图像处理(缩放、裁剪、旋转、转换、压缩、滤镜)
功能
| 工具 | 说明 |
|---|---|
vision_status |
服务器状态、当前 provider / 模型、配置告警 |
describe_image |
详细描述图像内容 |
ask_image |
针对图像的任意自然语言问答 |
ocr_image |
提取图中文字(OCR) |
process_image_tool |
本地处理:resize / crop / rotate / flip / convert / compress + 滤镜与亮度增强 |
快速开始
1. 克隆并安装
git clone <your-repo-url> vision-mcp
cd vision-mcp
python -m venv .venv
.venv\Scripts\pip install -r requirements.txt # Windows
# source .venv/bin/pip install -r requirements.txt # macOS / Linux
2. 配置 API Key
复制 .env.example 为 .env,填入后端 API key:
VISION_PROVIDER=openai # openai | gemini | anthropic
OPENAI_API_KEY=sk-...
# 若使用 OpenAI 兼容端点(智谱 / DeepSeek 等):
# OPENAI_BASE_URL=https://open.bigmodel.cn/api/paas/v4/
# OPENAI_VISION_MODEL=glm-4.6v-flash
只用你选择的 provider 对应的 key 即可。
3. 注册到 Claude Code
在 ~/.claude.json 或项目 .mcp.json 中添加:
{
"mcpServers": {
"vision-mcp": {
"command": "path/to/vision-mcp/.venv/Scripts/python.exe",
"args": ["-m", "vision_mcp.server"],
"cwd": "path/to/vision-mcp"
}
}
}
4. (可选)上传图片自动分析
将 image_attach_hook.py 作为 UserPromptSubmit hook 注册到 ~/.claude/settings.json:
{
"hooks": {
"UserPromptSubmit": [
{
"hooks": [
{
"type": "command",
"command": "path/to/vision-mcp/.venv/Scripts/python.exe path/to/vision-mcp/image_attach_hook.py",
"timeout": 90
}
]
}
]
}
}
上传图片时,hook 自动调用 describe_image + ocr_image,将视觉分析结果注入上下文——文本模型无需手动调用即可获得图片内容。
5. 手动测试
.venv\Scripts\python -m vision_mcp.server --help
# 或直接起 stdio 服务:
.venv\Scripts\python -m vision_mcp.server
图像路径约定
- 传绝对路径;或
- 传相对路径,并设置
VISION_IMAGE_BASE_DIR指向图片目录; - 也可以直接传
data:URI(例如处理工具产出的结果)。
技术架构
vision_mcp/
├── config.py # 多 provider 配置(OpenAI / Gemini / Anthropic)
├── vision.py # 视觉后端:图像理解、问答、OCR
├── image.py # 图片加载、预处理、格式转换
├── processor.py # 本地图像处理(resize/crop/rotate/滤镜等)
├── server.py # MCP 服务入口,注册 5 个工具
└── __init__.py
image_attach_hook.py # UserPromptSubmit hook:上传图片自动分析
run_vision_mcp.bat # Windows 启动脚本
e2e_test.py # 端到端测试脚本
requirements.txt # Python 依赖
许可
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.