VEverdun
audits

Dispute Event Audit

A writable staff decider disputes an overturn, once. Read-only accounts are refused. 403 unless the caller made the call; 409 unless the audit is overturned and not yet disputed.

post/audits/events/{event_id}/dispute

Request

  • Base URL
    https://api.verdun.ai (the document declares no server; this is the origin it was published from)
  • URL
    https://api.verdun.ai/audits/events/{event_id}/dispute
  • Auth
    HTTP bearer

Path parameters

event_idstring uuid required

Request body

notestring required

Why the overturn is wrong. Must not be blank; up to 2000 characters after stripping.

Response

Successful Response

event_idstring uuid required

The audited alarm.

evaluation_idstring uuid nullable

The audited evaluation; null once that evaluation or its deciding account is deleted.

operator_idstring uuid nullable

The decider; null once that account is deleted, and always null when the decider is the AI agent.

operator_namestring required

Snapshot of the decider's name; "Verdun Agent" when the decider is the AI agent.

decisionstring required

The decider's OperatorDisposition MEMBER NAME at audit time.

verdict'upheld' | 'overturned' required

The lead's ruling on the original decision.

correct_dispositionstring nullable

Overturned only: the disposition it should have been (member name).

reasonstring nullable

Overturned only: the auditor's reason.

auditor_idstring uuid nullable

The auditing lead; null once that account is deleted.

auditor_namestring required

Snapshot of the auditing lead's name.

ad_hocboolean required

True when the call was neither sampled nor rushed: a lead picked it themselves.

created_atstring date-time required

When the audit was made.

decider_is_agentboolean

True when the audited decision was the AI agent's close (not a human operator's). Nobody loses points on such an audit and nobody can dispute it.

pointsinteger required

-1000 (SCORE_OVERTURN) charged to the decider while an overturn is active (confirmed, not reversed on dispute), else 0. Always 0 when the decider is the AI agent.

catch_pointsinteger

Credited to a PEER auditor while their overturn is active, else 0 (a lead's overturn earns no catch): +1000 (SCORE_CATCH) for a human decider, +250 (SCORE_AGENT_CATCH) when the decider is the AI agent, 0 when auto_confirmed.

auto_confirmedboolean

A peer's overturn of an AI agent false alarm (Non-threatening -> False alarm): confirmed at once with no lead review (reviewed_by_id null) and worth 0 points. Only agent misses go to a lead and earn the agent catch.

auditor_is_leadboolean

Snapshot: the auditor was an owner or monitoring manager when they ruled (false = a peer operator).

review_status'pending' | 'confirmed' | 'rejected' nullable

Overturned only (null on an upheld audit). A lead's overturn is confirmed at once; a peer's is pending until a lead confirms or rejects it. Only a confirmed overturn costs points.

reviewed_by_idstring uuid nullable

The lead who confirmed or rejected the overturn (the auditor, for a lead's own); null while pending or once that account is deleted.

reviewed_by_namestring nullable

Snapshot of the reviewing lead's name.

reviewed_atstring date-time nullable

When the overturn was confirmed or rejected.

review_notestring nullable

The reviewing lead's optional note.

Changes

No recorded changes to this operation across all 2 revisions of this API.