1# Azure AI Content Safety — Python SDK Quick Reference
2 
3> Condensed from **azure-ai-contentsafety-py**. Full patterns (blocklist management, image analysis, 8-severity mode)
4> in the **azure-ai-contentsafety-py** plugin skill if installed.
5 
6## Install
7```bash
8pip install azure-ai-contentsafety
9```
10 
11## Quick Start
12```python
13from azure.ai.contentsafety import ContentSafetyClient, BlocklistClient
14from azure.ai.contentsafety.models import AnalyzeTextOptions, TextCategory
15client = ContentSafetyClient(endpoint=endpoint, credential=credential)
16```
17 
18## Non-Obvious Patterns
19- Two clients: `ContentSafetyClient` (analyze) and `BlocklistClient` (blocklist management)
20- Image from file: base64-encode bytes, pass via `ImageData(content=base64_str)`
21- 8-severity mode: `AnalyzeTextOptions(text=..., output_type=AnalyzeTextOutputType.EIGHT_SEVERITY_LEVELS)`
22- Blocklist analyze: `AnalyzeTextOptions(text=..., blocklist_names=[...], halt_on_blocklist_hit=True)`
23 
24## Best Practices
251. Use blocklists for domain-specific terms
262. Set severity thresholds appropriate for your use case
273. Handle multiple categories — content can be harmful in multiple ways
284. Use `halt_on_blocklist_hit` for immediate rejection
295. Log analysis results for audit and improvement
306. Consider 8-severity mode for finer-grained control
317. Pre-moderate AI outputs before showing to users
32