Task Graph Modeling and Constraint-Aware Scheduling for Reliable Large Language Model Agent Execution
Article Sidebar
Main Article Content
Abstract
This paper addresses the instability of large language model agents under real-world constraints, proposing an integrated approach of robust task decomposition and constraint-aware scheduling to improve the controllability, compliance, and stability of long-link interactive tasks. The method first structures high-level instructions into a task graph, explicitly modeling subtask dependencies, resource sharing, and executable conditions. Verifiable pre-checks, alternative paths, and fallback actions are configured for key steps to reduce error accumulation caused by incomplete information and environmental changes. Subsequently, during the scheduling phase, latency and risk are incorporated into a unified priority decision, dynamically selecting the execution order and allocating resources based on critical path identification and failure cost assessment. When tools become unavailable, constraints are triggered, or execution deviations occur, rapid recovery is achieved through local reordering and path switching, avoiding the high overhead of global rework. Comparative experimental results show that this method exhibits superior overall performance in key agent metrics such as task completion, tool invocation reliability, constraint violation control, interaction efficiency, and recovery capability. It can maintain a more stable end-to-end execution process under a limited budget and uncertain environments, providing a feasible agent execution paradigm for real-world business process automation and multi-tool collaboration scenarios.
Article Details
Issue
Section
References
[1]W. Huang, P. Abbeel, D. Pathak and I. Mordatch, "Language Models as Zero-Shot Planners: Extracting Actionable Knowledge for Embodied Agents," Proceedings of the International Conference on Machine Learning, pp. 9118-9147, 2022.
[2]Schick T, Dwivedi-Yu J, Dessì R, et al. Toolformer: Language models can teach themselves to use tools[J]. Advances in Neural Information Processing Systems, 2023, 36: 68539-68551.
[3]Karpas E, Abend O, Belinkov Y, et al. MRKL Systems: A modular, neuro-symbolic architecture that combines large language models, external knowledge sources and discrete reasoning[J]. arXiv preprint arXiv:2205.00445, 2022.
[4]Wang G, Xie Y, Jiang Y, et al. Voyager: An open-ended embodied agent with large language models[J]. arXiv preprint arXiv:2305.16291, 2023.
[5]M. Ahn, A. Brohan, N. Brown, Y. Chebotar, O. Cortes, B. David, A. Zeng et al., "Do as I can, not as I say: Grounding language in robotic affordances," arXiv preprint arXiv:2204.01691, 2022.
[6]J. Li, "LocateNet: Large Multimodal Models for Text-Guided Object Localization," 2024.
[7]S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. R. Narasimhan and Y. Cao, "ReAct: Synergizing reasoning and acting in language models," Proceedings of the International Conference on Learning Representations (ICLR), 2022.
[8]Y. Hu, "Autonomous Agent Architecture for Complex Tasks via Hierarchical Planning and Language Model Reasoning," 2024.
[9]N. Shinn, F. Cassano, A. Gopinath et al., "Reflexion: Language Agents with Verbal Reinforcement Learning," Advances in Neural Information Processing Systems, vol. 36, pp. 8634-8652, 2023.
[10]Q. Gan, "Large Language Model Framework for Multi-Document Financial Anomaly Detection in Intelligent Auditing via Semantic Mapping and Risk Reasoning," 2024.
[11]C. Hua, "A Semantic-Prior-Guided AI Framework for Collaborative Environment Understanding and Robust Agent Decision Making," 2024.
[12]Y. Li, "Task-aware Differential Privacy and Modular Structural Perturbation for Secure Fine-Tuning of Large Language Models," 2024.
[13]S. Yao, D. Yu, J. Zhao et al., "Tree of Thoughts: Deliberate Problem Solving with Large Language Models," Advances in Neural Information Processing Systems, vol. 36, pp. 11809-11822, 2023.
[14]Y. Deng, "Transfer Methods for Large Language Models in Low-Resource Text Generation Tasks," 2024.
[15]S. Yao, H. Chen, J. Yang et al., "WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agents," Advances in Neural Information Processing Systems, vol. 35, pp. 20744-20757, 2022.
[16]H. Liu, "Structural Regularization and Bias Mitigation in Low-Rank Fine-Tuning of LLMs," 2023.
[17]D. Zhou, N. Schärli, L. Hou, J. Wei, N. Scales, X. Wang et al., "Least-to-Most Prompting Enables Complex Reasoning in Large Language Models," Proceedings of the International Conference on Learning Representations (ICLR), 2023.
[18]Prasad A, Koller A, Hartmann M, et al. Adapt: As-needed decomposition and planning with language models[C]//Findings of the Association for Computational Linguistics: NAACL 2024. 2024: 4226-4252.
[19]X. Zhang, Y. Deng, Z. Ren, S. K. Ng, and T. S. Chua, “Ask-before-plan: Proactive language agents for real-world planning,” in Findings of the Association for Computational Linguistics: EMNLP 2024, Nov. 2024, pp. 10836-10863.
[20]S. Qiao, R. Fang, N. Zhang, Y. Zhu, X. Chen, S. Deng, et al., “Agent planning with world knowledge model,” Advances in Neural Information Processing Systems, vol. 37, pp. 114843-114871, 2024.
[21]X. Huang, W. Liu, X. Chen, X. Wang, H. Wang, D. Lian, et al., “Understanding the planning of LLM agents: A survey,” arXiv preprint arXiv:2402.02716, 2024.
[22]M. Hu, P. Zhao, C. Xu, Q. Sun, J. G. Lou, Q. Lin, et al., “AgentGen: Enhancing planning abilities for large language model based agent via environment and task generation,” in Proc. 31st ACM SIGKDD Conf. Knowledge Discovery and Data Mining (KDD), Jul. 2025, pp. 496-507.
[23]Yoran O, Amouyal S J, Malaviya C, et al. Assistantbench: Can web agents solve realistic and time-consuming tasks?[J]. arXiv preprint arXiv:2407.15711, 2024.
[24]A. Li, Y. Xie, S. Li, F. Tsung, B. Ding, and Y. Li, “Agent-oriented planning in multi-agent systems,” arXiv preprint arXiv:2410.02189, 2024.