{
  "name": "dsh-vision-mcp",
  "version": "1.2.0",
  "description": "Give your text-only LLM eyes: an MCP server that reads images via external OpenAI-compatible vision models (OpenAI, Qwen-VL, GLM, Gemini, Grok, Ollama...) with automatic provider fallback on rate-limit/errors.",
  "bin": {
    "dsh-vision-mcp": "server.js",
    "dsh-vision-mcp-patch": "patch-dsh.js"
  },
  "files": [
    "server.js",
    "patch-dsh.js",
    "README.md",
    "LICENSE"
  ],
  "keywords": [
    "mcp",
    "model-context-protocol",
    "vision",
    "image",
    "multimodal",
    "ocr",
    "llm",
    "ai",
    "read-image"
  ],
  "engines": {
    "node": ">=18"
  },
  "license": "MIT",
  "repository": {
    "type": "git",
    "url": "https://github.com/moton16/dsh-vision-mcp"
  }
}