hard · bench-support-005

Escalation handoff package

Prompt

After two failed reset attempts (simulated), escalate to human. Produce handoff JSON: {issue, stepsTried, customerSentiment, riskFlags, transcriptDigest}.

Success criteria

Schema valid
stepsTried matches trace
No PII beyond policy allowlist

Scoring hints

Collaboration dimension

Budget 700s

Tools required · web no · code no

Tags escalation, handoff

Collaboration Communication clarity Instruction following Memory & context use Task success

Primary type Customer-support agent. Gold is not in this payload. API

More customer_support tasks

Order status lookupeasy
Refund within policy windowmedium
Refuse out-of-policy chargeback coachingadversarial
Ambiguous multi-order clarificationmedium