{"ID":2827438,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2512.16279","arxiv_id":"2512.16279","title":"QuadSentinel: Sequent Safety for Machine-Checkable Control in Multi-agent Systems","abstract":"Safety risks arise as large language model-based agents solve complex tasks with tools, multi-step plans, and inter-agent messages. However, deployer-written policies in natural language are ambiguous and context dependent, so they map poorly to machine-checkable rules, and runtime enforcement is unreliable. Expressing safety policies as sequents, we propose \\textsc{QuadSentinel}, a four-agent guard (state tracker, policy verifier, threat watcher, and referee) that compiles these policies into machine-checkable rules built from predicates over observable state and enforces them online. Referee logic plus an efficient top-$k$ predicate updater keeps costs low by prioritizing checks and resolving conflicts hierarchically. Measured on ST-WebAgentBench (ICML CUA~'25) and AgentHarm (ICLR~'25), \\textsc{QuadSentinel} improves guardrail accuracy and rule recall while reducing false positives. Against single-agent baselines such as ShieldAgent (ICML~'25), it yields better overall safety control. Near-term deployments can adopt this pattern without modifying core agents by keeping policies separate and machine-checkable. Our code will be made publicly available at https://github.com/yyiliu/QuadSentinel.","short_abstract":"Safety risks arise as large language model-based agents solve complex tasks with tools, multi-step plans, and inter-agent messages. However, deployer-written policies in natural language are ambiguous and context dependent, so they map poorly to machine-checkable rules, and runtime enforcement is unreliable. Expressing...","url_abs":"https://arxiv.org/abs/2512.16279","url_pdf":"https://arxiv.org/pdf/2512.16279v1","authors":"[\"Yiliu Yang\",\"Yilei Jiang\",\"Qunzhong Wang\",\"Yingshui Tan\",\"Xiaoyong Zhu\",\"Sherman S. M. Chow\",\"Bo Zheng\",\"Xiangyu Yue\"]","published":"2025-12-18T07:58:40Z","proceeding":"cs.AI","tasks":"[\"cs.AI\",\"cs.CL\"]","methods":"[\"Language Model\"]","has_code":false,"code_links":[{"ID":605807,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2827438,"paper_url":"https://arxiv.org/abs/2512.16279","paper_title":"QuadSentinel: Sequent Safety for Machine-Checkable Control in Multi-agent Systems","repo_url":"https://github.com/yyiliu/QuadSentinel","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}