Toxicity
Schema-driven detector documentation.
TOXICactiveP16 params2 examples
Detector Metadata
Capability catalog entry from
all_detectors.json.Categories
CONTENT
Supported Asset Types
TXTTABLEURL
Recommended Model
detoxifyParameters
Configuration parameters for the Toxicity detector. Shared from `ContentDetectorConfig`.
| Parameter | Type | Required | Description | Default | Constraints |
|---|---|---|---|---|---|
| enabled_patterns | array | No | Specific content types to detect | — | — |
| enabled_patterns[] | enum | No | Content detector pattern types Allowed values: toxicity, severe_toxicity, obscene, threat, insult, identity_attack, nsfw, nsfw_explicit | — | — |
| severity_threshold | enum | null | No | Minimum severity to report | null | — |
| confidence_threshold | number | No | Minimum confidence to report (0-1) | 0.7 | min 0, max 1 |
| max_findings | integer | null | No | Maximum number of findings to return | null | — |
| model_name | enum | No | Detoxify model variant Allowed values: original, unbiased, multilingual | original | — |