neo_vision_image_to_json

v1.0.0

Extracts structured JSON from an image using a Vision LLM.

$0.055 / call (Ultra Tier)

Updated: May 02, 2026

Overview

An advanced multimodal extraction tool. It uses Vision LLM capabilities (Qwen3-VL) to analyze images, screenshots, and diagrams, extracting the visual data and mapping it directly into a structured JSON schema you provide.

Example Input

JSON payload sent to this tool:

{
  "image_base64": "/9j/4AAQSkZJRgABAQAAAQABAAD...",
  "resource_uri": "neonia://resource/c4d12c82-...",
  "prompt": "Analyze this image and extract the cat details.",
  "schema": {
    "type": "object",
    "properties": {
      "has_cat": { "type": "boolean", "description": "Is there a cat in the image?" },
      "cat_color": { "type": "string", "description": "The color of the cat, if present." }
    },
    "required": ["has_cat", "cat_color"],
    "additionalProperties": false
  }
}

Example Output

Formatted JSON response returned by this tool:

{
  "has_cat": true,
  "cat_color": "orange"
}

Setup Configuration

Select MCP Client

Add the following configuration to your MCP general settings or mcp_config.json:

{
  "mcpServers": {
    "neonia": {
      "serverUrl": "https://mcp.neonia.io/mcp?tools=neo_vision_image_to_json",
      "headers": {
        "Authorization": "Bearer API_KEY"
      }
    }
  }
}