cff-version: 1.2.0
message: "If you use WildClawBench in your research, please cite it using the metadata below."
title: "WildClawBench"
type: software
license: MIT
abstract: "WildClawBench is an in-the-wild benchmark for AI agents in the OpenClaw environment, designed to evaluate end-to-end agent performance across 60 tasks spanning productivity, coding, social interaction, search, creative synthesis, and safety alignment."
repository-code: "https://github.com/internlm/WildClawBench"
url: "https://github.com/internlm/WildClawBench"
keywords:
  - AI agents
  - benchmark
  - evaluation
  - OpenClaw
  - multimodal agents
authors:
  - family-names: "Ding"
    given-names: "Shuangrui"
  - family-names: "Dai"
    given-names: "Xuanlang"
  - family-names: "Xing"
    given-names: "Long"
  - family-names: "Ding"
    given-names: "Shengyuan"
  - family-names: "Liu"
    given-names: "Ziyu"
  - family-names: "Yang"
    given-names: "Jingyi"
  - family-names: "Yang"
    given-names: "Penghui"
  - family-names: "Zhang"
    given-names: "Zhixiong"
  - family-names: "Wei"
    given-names: "Xilin"
  - family-names: "Fang"
    given-names: "Xinyu"
  - family-names: "Ma"
    given-names: "Yubo"
  - family-names: "Duan"
    given-names: "Haodong"
  - family-names: "Shao"
    given-names: "Jing"
  - family-names: "Wang"
    given-names: "Jiaqi"
  - family-names: "Lin"
    given-names: "Dahua"
  - family-names: "Chen"
    given-names: "Kai"
  - family-names: "Zang"
    given-names: "Yuhang"