Primary sourceViking
View sourceEvent ·Multi-source
OpenClaw 开源 test audit skill:让 agent 判断哪些测试该删
OpenClaw 公开了自己使用的 test audit skill,任务是告诉 agent 什么测试应该加、什么测试是垃圾不该加。agent 在写测试前要先回答四个问题,判断通过才动手。此前 Viking 提到 OpenClaw 删掉了约 40 万行自己的测试代码,代码覆盖率几乎没掉,原因是现在的模型爱为测试而测试,写十个 case 去判断常量里的字符串。作者还建议给 agent 一个更狠的目标,比如删掉最没用的 20% 测试同时覆盖率波动控制在 2% 以内,只说“清理一下”它会很早就停下。
Current read
OpenClaw 把自家删了 40 万行测试的 skill 开源了,agent 写测试前先过四个问题,做 AI 编程的可以直接抄这个工作流。
8 sources63° AI heatLast updated 2026/9/25 13:06:32
Evidence
8 sourcesRelated source 1Decoder
View sourceRelated source 2IT之家
View sourceRelated source 3openclaw
View sourceRelated source 4theverge
View sourceRelated source 5arXiv: DeepSeek
View sourceRelated source 6rohanpaul_ai
View sourceRelated source 7王树义
View sourceConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.