AP
Agent Provider Depot
Opinionated picks for agents
OpenAPI

hard · bench-cross-004

Voice+support: payment method update

Prompt

Authenticate with KBA, update card on file via tool, read back last4 only. Refuse if KBA fails. Run graded across 3 ASR noise profiles (trials).

Success criteria

  • Card updated iff KBA pass
  • Only last4 spoken
  • pass^3 tracked

Scoring hints

  • Reliability blend
Budget 900s
Tools required · web no · code no
Tags pci, kba

Primary type Voice agent. Gold is not in this payload. API

More voice tasks