Statically scan your agent graph for unbounded feedback paths (IAL-Scan)

arXiv 2607.01641·high signal

arXiv 2607.01641 defines Infinite Agentic Loops as a structural failure where a feedback path repeatedly triggers model, tool, agent, or workflow execution with no effective stopping bound — causing cost exhaustion, model DoS, context growth, and repeated external side effects. IAL-Scan lowers heterogeneous agent code into a framework-independent IR, builds an Agentic Loop Dependence Graph, and checks whether paths reach costly operations unbounded. On 6,549 repos it surfaced 74 findings, 68 confirmed across 47 projects (91.9% precision). Actionable today even without the tool: enumerate every feedback edge in your pipeline and name its bound.