What categories of content can this skill detect?

The Azure AI Content Safety skill detects four main harm categories: Hate, Sexual, Violence, and Self-Harm.

What authentication methods are supported?

It supports both standard Azure Key Credentials and Entra ID via DefaultAzureCredential for secure access.

Can I use custom blocked words?

Yes, the skill includes full support for managing custom blocklists and halting analysis when a blocked term is hit.

Does it support image moderation?

Yes, it can analyze images from local base64-encoded files or remote blob URLs to detect harmful visual content.

How detailed are the severity scores?

By default, it uses a 4-level scale (0, 2, 4, 6), but it can be configured for a finer 8-level scale (0-7) for more precise control.

Azure AI Content Safety for Python

Name: Azure AI Content Safety for Python
Author: microsoft

bymicrosoft

•

1,777

•

安全与测试

Moderates user-generated and AI-generated content by detecting harmful text and images with multi-severity classification.

This skill integrates the Azure AI Content Safety SDK for Python into Claude's workflow, enabling automated moderation of text and visual media. It allows developers to identify and filter categories such as hate speech, self-harm, sexual content, and violence using customizable severity thresholds. With support for both built-in analysis models and custom blocklists, it provides a robust framework for ensuring application safety, compliance, and user protection by grounding Claude in Azure's enterprise-grade moderation tools.

主要功能

01Support for both API Key and Entra ID (Azure AD) authentication methods

02Custom text blocklist management for domain-specific terminology filtering

03Flexible image processing for local files and remote blob URLs

04Multi-category text and image analysis for hate, violence, and self-harm detection

051,777 GitHub stars

06Granular severity scoring with support for 4-level and 8-level severity scales

使用场景

01Pre-moderating AI-generated responses to ensure safety and brand alignment

02Automating moderation of user comments and forum posts in Python applications

03Building custom content filtering systems with industry-specific blocklists

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add microsoft/skills azure-ai-contentsafety-py

For use in Claude.ai and ChatGPT

Download Skill