Skip to main content

Promptinsight

A prompt security insight event.

Properties

NameTypeDescriptionNotes
timestampdatetimeEvent time in UTC.[optional]
userstrUser identifier or display name.[optional]
agentstrThe AI agent that processed the prompt.[optional]
policy_decisionEnum [ 'ALLOWED', 'REDACTED' ]The policy decision applied to the prompt.[optional]
categoryEnum [ 'ANOMALIES', 'DATA_UPLOADS', 'MCP_TOOL_CALLS' ]The category of the prompt security finding.[optional]
severityEnum [ 'LOW', 'MEDIUM', 'HIGH', 'CRITICAL' ]The severity of the prompt security finding.[optional]
reasonstrHuman-readable or structured reason for the policy decision.[optional]
rulestrThe rule that matched the prompt.[optional]
policystrThe policy that matched the prompt.[optional]
}

Example

from sailpoint.prompt_insights.models.promptinsight import Promptinsight

promptinsight = Promptinsight(
timestamp='2026-05-14T10:01:02.345Z',
user='john.doe@mail.com',
agent='ChatGPT',
policy_decision='REDACTED',
category='ANOMALIES',
severity='HIGH',
reason='Policy matched suspicious system override pattern',
rule='Prompt Guard - Clean',
policy='PG-INFO:None'
)

[Back to top]