Skip to content
Documentation

NERV Agent interface

NERV Agent is the conversational control surface for an investigation. It turns natural-language objectives into visible steps and keeps replies connected to the active case.

Main flow

  1. Open case
  2. State outcome
  3. Resolve questions
  4. Follow progress
  5. Review result
  1. Open a case and describe the intended outcome.
  2. Review any questions, candidate identities or proposed actions.
  3. Confirm the correct subject or scope.
  4. Follow progress and inspect the evidence or artefacts returned by the agent.
  5. Continue, correct or stop the work without losing the case context.

Conversation controls

The interface can reduce repeated explanations when a thread becomes long and can clear the visible conversation when a fresh start is required. Clearing chat does not silently rewrite the underlying case record.

Good practice

State the target, jurisdiction, time range and desired output. Resolve ambiguity before requesting a deep crawl, and treat generated conclusions as analysis that must remain linked to its sources.