name: WildClawBench
description: Hard, practical, end-to-end evaluation for AI agents — in the wild.
evaluation_framework: wildclawbench
tasks:
  - id: overall
    name: Overall
    config: default
    split: test
  - id: avg_time
    name: Avg Time (min)
    config: default
    split: test
  - id: avg_cost
    name: Avg Cost ($)
    config: default
    split: test