Privent icon

Privent

Privent DLP: session-scoped tokenization, detokenization, risk scoring, audit events and agent handoffs for AI agent workflows.

Actions6

Audit → Emit

AI-generated

Summary

Emit an audit event for each input item to Privent, with event details and optional metadata. This operation is available in API Key authentication mode; LLM Call events include model and token counts for backend cost calculation.

Inputs

  • Authentication (required) — Choose API Key to use this operation. The other declared modes are Tokenless and Local, but the audit inputs and event emission are scoped to API Key mode.
  • Session ID (required) — Session ID from an upstream Privent Session node.
  • Trace ID — Correlation ID linking the event to a session trace; defaults from an upstream Privent Session node.
  • Agent Name — Logical agent identifier to record on the event; defaults from an upstream Privent Session node.
  • Event Type — Select Egress, Error, LLM Call, or Policy Decision. Defaults to LLM Call; that event type triggers backend cost calculation from ModelPricing.
  • LLM Model — For LLM Call events, select a priced model or enter a provider|model identifier.
  • Prompt Tokens — For LLM Call events, expression for the prompt token count; defaults to usage.prompt_tokens on the incoming item.
  • Completion Tokens — For LLM Call events, expression for the completion token count; defaults to usage.completion_tokens on the incoming item.
  • Extra Metadata (JSON) — Optional JSON merged into event metadata; use it for event-specific fields such as policy decision rationale.

Output shape

A list of items containing the audit result for each input item.

The operation sends audit events to the Privent backend using the configured API key. Exact response fields are not established by the supplied operation source view.

Links

Discussion