Actions21
- Azure OpenAI Actions
- Batch API Actions
- Content Safety Actions
- Document Intelligence Actions
- Language Actions
- Speech Actions
- Translator Actions
- Vision Actions
Content Safety → Analyze Text
AI-generatedSummary
Analyze the supplied text for hate, violence, self-harm, and sexual content, then mark it unsafe if any category’s severity meets the configured threshold.
Inputs
- Authentication — Choose API Key (default) or Entra ID (OAuth2) and configure the matching Azure credentials.
- Text (required) — Text to analyze. The operation errors if this value is empty.
- Safety Threshold — Severity cutoff from 0 to 6, default 2. A category at or above this score makes the result unsafe; lower thresholds are stricter.
Output shape
One JSON result per input item, containing safe and categories fields.
categories contains severity scores for the categories returned by Azure: hate, violence, selfHarm, and sexual. The node requests all four categories using FourSeverityLevels; it returns no raw response field.
Examples
Example 1: Analyze a text field from an incoming item
Set Text to an expression referencing the incoming text and choose a Safety Threshold, such as the default 2.