从 Agent 轨迹构建自动机：失败与下一步预测

View PDF HTML (experimental)

Abstract:LLM-based agents execute multi-step tasks, but their behavioral structure remains opaque: long unstructured traces resist the safety auditing and runtime monitoring that deployment requires. Existing approaches operate per-trace or success-only, so they miss the cross-run topology that links next-step and failure prediction. To recover that shared structure, we collapse an entire trace corpus into a single, compact finite-state machine (FSM) that serves as a structural substrate for the otherwise unpredictable behavior of LLM agents. Across twelve public datasets, the FSMs are compact (7-43 states), replay held-out data at >=0.997 fitness with near-identical topology across splits, and build in milliseconds. This substrate addresses both prediction goals. For next-step prediction, FSM-state context outperforms Agent Workflow Memory on every ground-truth-matched dataset. For failure prediction, per-state behavioral features reach held-out AUROC up to 0.94, and an online monitor ranks failing runs above passing ones from a partial trace, triggering early stopping well before completion. Behavioral topology thus appears shaped more by the deployment harness than by the LLM, providing a model-agnostic structural primitive for safety auditing and runtime monitoring.

Comments:	Accepted at NeurIPS 2026. 43 pages, 17 figures
Subjects:	Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Cite as:	arXiv:2608.23670 [cs.AI]
	(or arXiv:2608.23670v2 [cs.AI] for this version)
	https://doi.org/10.48550/arXiv.2608.23670 arXiv-issued DOI via DataCite

Submission history

From: Seonglae Cho [view email]
[v1] Mon, 24 Aug 2026 17:58:01 UTC (1,199 KB)
[v2] Thu, 12 Nov 2026 06:59:30 UTC (4,248 KB)