---

name: image-audit

description: Detects adult, political and violent content in images via automated batch auditing. Compresses large images, calls MCP audit service, and outputs results as a table. Use when auditing images, checking image content, scanning photos for inappropriate material, or when the user says audit images, review pictures, check content, or image moderation.

license: MIT

compatibility: Requires node npm and nx-mcp-audit MCP service with NX_API_KEY configured

metadata:

  author: zhjian_2026

  version: 1.0.1

  tags:

    - image-audit

    - content-moderation

    - batch-processing

---



# Image Content Moderation



Audit images for adult, political, and violent content using the nx-mcp-audit MCP service.



## Setup



Configure the MCP server and API key in `.mcp.json` (not `settings.json` — it doesn't support the `mcpServers` field):



```json

{

  "mcpServers": {

    "nx-mcp-audit": {

      "type": "url",

      "url": "https://mcp.api-inference.modelscope.net/da16b3f65bdb4e/mcp",

      "env": {

        "NX_API_KEY": "your-api-key-here"

      }

    }

  }

}

```



Restart Claude Code after configuring.



> **No API Key?** Contact WeChat `zhjian_2026` to get one.



## Usage



Say "audit these images" and the skill will:



1. Collect images from a folder path or URL

2. Compress large images to fit gateway limits

3. Call the nx_img_audit MCP tool for each image

4. Summarize results in a table with pass/block/fail status

5. Provide recommendations for blocked images



## Response Fields



- safe: true for pass, false for blocked

- source: audit engine (e.g. wechat)

- summary: aggregate stats (total, pass, block, error)