返回全部动态
SkillTrace:LLM智能体技能复用的多轨迹溯源审计框架
原标题:SkillTrace: Multi-Trace Provenance Auditing for LLM-Agent Skill Reuse
AI 摘要
arXiv 论文提出 SkillTrace,一个用于 LLM 智能体技能复用审计的多轨迹溯源框架。该框架提取表达、实现和操作三条轨迹,其中操作轨迹用技能操作图表示,并利用 LLM 辅助提取。在 SkillTrace-Bench 基准上达到 AUROC 0.938 和 F1 0.898,并在 36,446 个技能的真实审计中优于基线方法。
以上摘要由 AI 生成,可能存在误差。事实请以原文为准。
正文节选
Computer Science > Artificial Intelligence Title:SkillTrace: Multi-Trace Provenance Auditing for LLM-Agent Skill Reuse View PDF HTML (experimental) Abstract:LLM-agent ecosystems are rapidly growing around reusable skills: mixed-modality packages of metadata, natural-language instructions, code, tools, references, and operational workflows. As skills become marketplace artifacts, auditing their reuse is no longer the same problem as ordinary code clone detection. Existing detectors ta
发布时间:2026-08-07 12:00
抓取时间:2026-08-07 14:21
来源机构:arXiv