Vision MCP

让没有原生识图能力的模型（如 DeepSeek）也能"看图"——通过调用外部视觉 API 获取图片的文字描述。

Provide image recognition for models without native vision (e.g. DeepSeek) — by calling an external visual API.

# 安装依赖（本地开发） npm install npm run build # 或直接通过 npx 运行（无需安装） npx -y @guorui99/vision-mcp

配置 MCP（在 Claude Desktop / claude.json 中添加）:

{ "mcpServers": { "vision": { "command": "npx", "args": ["-y", "@guorui99/vision-mcp"] } } }

vision_update_config api_key="你的API Key"

VISION_API_KEY=你的Key npx -y @guorui99/vision-mcp

jpg, jpeg, png, gif, webp, bmp（单次最多 5 张，单文件最大 20MB）
- 读取图片 → base64 编码
- 调用 OpenAI 兼容视觉 API
- 返回文字描述

# 单张图片 node scripts/vision.cjs photo.jpg "描述这张图片" # 网络图片 node scripts/vision.cjs --url https://example.com/img.png "这是什么？" # 多张图片 node scripts/vision.cjs img1.jpg img2.jpg "比较这两张图" # 配置 node scripts/vision.cjs --setup node scripts/vision.cjs --config

vision-mcp/ ├── src/ # MCP Server 源码 (TypeScript) ├── dist/ # 编译产物 ├── scripts/vision.cjs # CLI 工具 (legacy) ├── config.json # 共享配置文件 ├── package.json └── tsconfig.json

- Node.js >= 18
- 视觉 API 的 Key（阿里云百炼 / OpenAI 等）

# Local development npm install npm run build # Or run via npx directly (no install needed) npx -y @guorui99/vision-mcp

{ "mcpServers": { "vision": { "command": "npx", "args": ["-y", "@guorui99/vision-mcp"] } } }

vision_update_config api_key="your-api-key"

Or use environment variables (no disk writes):

VISION_API_KEY=your-key npx -y @guorui99/vision-mcp

jpg, jpeg, png, gif, webp, bmp (max 5 images per request, max 20MB per file)
- Read image → base64 encode
- Call OpenAI-compatible vision API
- Return text description

# Single image node scripts/vision.cjs photo.jpg "Describe this image" # Remote image node scripts/vision.cjs --url https://example.com/img.png "What is this?" # Multiple images node scripts/vision.cjs img1.jpg img2.jpg "Compare these two" # Setup / Config node scripts/vision.cjs --setup node scripts/vision.cjs --config

vision-mcp/ ├── src/ # MCP Server source (TypeScript) ├── dist/ # Compiled output ├── scripts/vision.cjs # CLI tool (legacy) ├── config.json # Shared config file ├── package.json └── tsconfig.json

- Node.js >= 18
- API key for a vision service (Alibaba Cloud Bailian / OpenAI / etc.)

Search global news using natural language. Webz.io News Search API returns the most relevant articles and content, with filters for source, country, language, date, sentiment, and category.

Generate images using the Together AI API. Supports custom aspect ratios, save paths, and batch generation.

Identity infrastructure for the AI economy. 33-trait psychometric engine — belonging probability, trait vectors, attunement depth. Remote streamable-HTTP. Free tier: 16 tools, 10 req/min.

An unofficial MCP server for the Artificial Analysis API

Comprehensive astrology MCP backed by the AstroWay Calculation API — natal, synastry, transits, Vedic dashas (Vimshottari/Yogini/Ashtottari/Kalachakra), Tarot (Rider-Waite-Smith / Marseille / Lenormand), Numerology, Human Design, AI horoscopes. Sub-arcsecond Swiss Ephemeris precision. 10 000 free credits/month, no card required. Install: npx @astroway/mcp

Access 30+ AI models in Claude Code with zero API keys. One wallet, pay-per-request.

An image generation server that connects to a local ComfyUI instance via its API, supporting dynamic workflows.

Token cost math for LLM API calls: 69 models across 17 providers, prices verified by ComparEdge. Free, no API key.

A Model Context Protocol (MCP) server that provides access to multiple Large Language Model (LLM) APIs including ChatGPT, Claude, Gemini, and DeepSeek.

Provides a full suite of AI tools via DeepInfra’s OpenAI-compatible API, including image generation, text processing, embeddings, and speech recognition.