Moderation
Rules that inspect the content flowing through your agents, and the violations they raise, from a keyword or regex match to a built-in detector.
The Moderation pillar lets you define rules over the content your agents produce and consume, and review the violations those rules raise.
A rule
| Field | Values |
|---|---|
name | A label you will recognise. |
detector | A built-in detector name, or regex / keyword. |
pattern | The pattern the detector matches on, for regex and keyword. |
action | allow, redact, block or flag. |
redact masks the matched content, block refuses it, and flag lets it
through but records a violation for review. Create and list rules at
/moderation/rules.
Violations
Every match a rule raises is recorded as a violation, browsable at
/moderation/violations. Each one carries the rule that fired, the action taken
and the trace it came from, so a flagged run is one click from its full trace in
Observe.
Related
Alerts
Fire a notification when a finished trace crosses a threshold you set, delivered to email or an existing webhook endpoint.
Govern
The controls that bound and gate LLM traffic — governance policies that rewrite or reject non-compliant requests, spend policies, quotas, model allow-lists, API keys, members and SSO — plus environments, the telemetry policy and export destinations.