Translation in progress Some entries still include text in the original language.
Failure-First·Indexed by this site · Originally published
Revelio:面向仓库级代码、成本高效的智能体内存安全漏洞检测
[Daily Paper] Revelio: Cost-Efficient Agentic Memory Safety Vulnerability Detection For Repository-Scale Code
AI summary
Revelio 用 LLM 引导的污点分析把内存安全审查扩展到数百万行代码的仓库,优先排查可从不可信输入到达、且更可能含内存错误的代码路径。其 LLM 引导的优先级排序将待审路径集缩小约 12 倍并保持高召回率;结构化记忆记录已分析区域、已确认漏洞与未审查路径,支持随代码库演进的增量分析。
Read originalfailurefirst.org