Newest papers

AgentTriage: Replayable Security Diagnosis Across Heterogeneous Agent Runtimes

Jie Ying, Xutao Zhang, Ningyuan Chen, Mengce Zheng, Jun Li, Zhongjie Zhu, Tiantian Zhu · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.30738 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Quando os agentes de uso pessoal escrevem arquivos, executam código ou enviam dados, suas respostas finais não explicam os eventos de segurança resultantes, e os tempos de execução heterogêneos expõem a evidência relevante através de interfaces incompatíveis. Nós apresentamos o AgentTriage, um contrato executável para o diagnóstico de segurança reproduzível em tempos de execução heterogêneos. O contrato admite conclusões cujas reivindicações citam eventos de execução canônicos e requer a abstenção quando a evidência obrigatória está ausente. O Runtime sobrepõe as mesmas observações nativas, enquanto um pipeline determinista junta invocação e ciclos de vida de resultados, constrói um gráfico de proveniência tipificado, seleciona uma faixa de decisão de política e vincula as reivindica

Machine translation (Meta M2M100); the English original follows.

Abstract

When tool-use agents write files, execute code, or send data, their final responses do not explain the resulting security events, and heterogeneous runtimes expose the relevant evidence through incompatible interfaces. We present AgentTriage, an executable contract for replayable security diagnosis across heterogeneous runtimes. The contract admits conclusions whose claims cite canonical runtime events and requires abstention when mandatory evidence is missing. Runtime overlays normalize native observations, while a deterministic pipeline joins invocation and result lifecycles, constructs a typed provenance graph, selects a policy-decisive slice, and binds claims to recorded evidence. Given the same normalized trace, overlay version, and pre-run policy oracle, the diagnosis can be recomputed and its claims re-checked without language-model judgment. Across five runtimes, one contract captures all 159 applicable executions, with an audit of native and synthesized identifiers. On 120 adjudicated cases, AgentTriage achieves 0.858 grounded joint diagnosis, 0.966 required-evidence coverage, and a 0.036 unsupported-claim rate, improving joint diagnosis by 0.175 (95% CI [0.092, 0.258]) over the strongest same-schema baseline. These results quantify the precision of replayable diagnosis and the observation boundaries that require abstention. We release an open-source implementation and selected sanitized evidence at https://github.com/AsnGs/agenttriage-open-source.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Executable, Joins, Pipeline (software), Audit, Exploit

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange