One Gate Is Not Enough: Composing Stateful Pre-Action Controls for Agentic AI
Source
arxiv.org
Author
Gaston Besanson
Date
Why it matters
If your AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → runs multiple pre-action gates, a fix applied by one can silently invalidate another's approval — and remediation order changes the outcome, so it belongs in your control-plane spec rather than in whatever order the code happens to run.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
articleWhen Agent Governance HelpsMichael Ray Johnson, Linda Naimi
articleAssurance Envelopes for Autonomous Coding Agents: Minimum-Cost Evidence for Software ChangeAnjan Goswami
articleOpenAgentFlow: Enabling System-Wide Safety Boundaries for Heterogeneous AI Agent FleetsDongsheng Chen, Xiangyu Zhao, Xin Yao, Xuetao Wei