Skip to main content
POST
Pre-check content against moderation policy

Authorizations

Authorization
string
header
required

API key or access token

Body

application/json
text
string
required

The text to evaluate. Up to 10,000 characters; longer text is truncated server-side.

Required string length: 1 - 10000
direction
enum<string>
default:input

Which pattern set to apply. "input" runs the brief/jailbreak/policy patterns; "output" runs the LLM-output (refusal/identity/PII) patterns.

Available options:
input,
output
surface
string
default:moderation.check

Optional surface label for metric attribution. Defaults to "moderation.check".

Required string length: 1 - 100

Response

Pre-check content against moderation policy

passed
boolean
required

True when no moderation patterns matched.

wouldBlock
boolean
required

True when at least one finding would trigger a 422 if this content were submitted to a blocking surface.

findings
object[]
required

All matching findings. Empty when passed is true. Both blocking and non-blocking findings are reported so callers can observe medium-severity signals.