TRtraversal

Investigation Follow Up V2

Changed on

V2 investigation follow-up endpoint that accepts thinking_mode parameter.

The thinking_mode parameter determines how the follow-up is conducted:

  • AUTO: Uses message classifier to determine RCA vs CHAT investigation
  • DEEP: Forces RCA (root cause analysis) investigation
  • FAST: Forces CHAT (exploratory) investigation
post/v2/investigation/follow-up

Request

  • Base URL
    https://api.traversal.com (the document declares no server; this is the origin it was published from)
  • URL
    https://api.traversal.com/v2/investigation/follow-up
  • Auth
    none declared

Request body

session_idstring required
user_uuidstring nullable
user_messagestring required
config_namestring required
mode'rca' | 'chat' | 'triage' | 'alert_story'
thinking_mode'auto' | 'deep' | 'fast' | 'lightning' | 'ultra'

Determines the investigation depth and approach.

  • AUTO: Use message classifier to determine RCA vs CHAT
  • DEEP: Force RCA (root cause analysis) investigation
  • FAST: Force CHAT (exploratory) investigation
  • LIGHTNING: Fast, tool-rich investigation with tight iteration cap. Uses the union of all tools from fast + deep configs with a speed-focused prompt.
  • ULTRA: Same agent config as DEEP. The client-side trigger for first-turn ensembling. A no-op on follow-ups.

Response

Successful Response

session_idstring required
system_chat_idstring required
user_chat_idstring required

Changes

  • c5e1ea955e81Revision changes
    • info

      added the new ultra enum value to the request property /