AI Governance Audit Chain(AI 留痕与门禁)

Give your AI an identity, an evidence trail, and a gate. Every task gets a registry entry (who, what, on what authority), every step leaves a verifiable record (what changed, where the proof is, how to recompute it), and every high-risk action must pass a gate first. Use for AI overreach, runaway agents, unsourced AI conclusions, automation needing human approval, audit trails, accountability chains, and prompt-injection defense. 「AI 改了我的文件,怎么查是谁改的?」「AI 自动帮我发邮件、发帖,怎么拦住?」「让 AI 干活,怎么留痕?」 装上它,AI 会先给自己立三条规矩:每个任务建籍(谁在干、干什么、依据什么)、每步操作留证(改了什么、证据在哪、怎么复算)、 关键动作设门禁(对外发送、删除覆盖、付款下单之前先过闸)。适用于 AI 越权操作、智能体乱来、AI 结论没标来源、 自动化流程要人工审批、合规审计要留痕、事后追责要证据链、提示词注入防范、AI 可靠性提升。

Install

openclaw skills install @zhaoxinghua09-cell/ai-governance-audit-chain