Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Detects adult and inappropriate content in images using a local ONNX model, requiring no external API. Classifies content across categories including porn, sexy, hentai, and drawings with confidence scores from 0-100%. Supports context-aware exceptions for medical and art imagery.
Runs entirely on a local ONNX model, so images never leave the user's environment and no API key or external service is required.
When to use it
- Blocking explicit uploads on a user-generated content platform
- Moderating profile photos before they go public on a social app
- Enforcing child-safe mode in a family-oriented web tool
- Pre-filtering image URLs before rendering in a browser extension
- Flagging adult content in automated screenshot or web-scraping pipelines
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: scan this image URL and tell me if it should be blocked
- 1Import detectNSFW from @clawhub/nsfw-detector
- 2Call detectNSFW with the provided image URL
- 3Read the returned confidence score and category breakdown
- 4Compare confidence against the configured threshold
- 5Return the isNSFW flag and recommended action (block, warn, or allow)
{ isNSFW: true, confidence: 0.92, categories: { porn: 0.92, sexy: 0.15 }, action: 'block' }