Enhancing LLM-Based Agents via Global Planning and Hierarchical Execution

TL;DR

GoalAct tackles core limitations of LLM-based agents in planning and execution by coupling a continuously updated global plan with a hierarchical execution strategy built on high-level skills. By simplifying planning to high-level skills and tightly integrating planning with execution, GoalAct achieves coherent, executable strategies that scale to diverse tasks. On LegalAgentBench, GoalAct delivers state-of-the-art performance with an average improvement of 12.22% in success rate across multiple LLMs. The work highlights the value of planning-execution coupling and leaves room for future enhancements through memory and reflective modules to further boost robustness and generalization.

Abstract

Intelligent agent systems based on Large Language Models (LLMs) have shown great potential in real-world applications. However, existing agent frameworks still face critical limitations in task planning and execution, restricting their effectiveness and generalizability. Specifically, current planning methods often lack clear global goals, leading agents to get stuck in local branches, or produce non-executable plans. Meanwhile, existing execution mechanisms struggle to balance complexity and stability, and their limited action space restricts their ability to handle diverse real-world tasks. To address these limitations, we propose GoalAct, a novel agent framework that introduces a continuously updated global planning mechanism and integrates a hierarchical execution strategy. GoalAct decomposes task execution into high-level skills, including searching, coding, writing and more, thereby reducing planning complexity while enhancing the agents' adaptability across diverse task scenarios. We evaluate GoalAct on LegalAgentBench, a benchmark with multiple types of legal tasks that require the use of multiple types of tools. Experimental results demonstrate that GoalAct achieves state-of-the-art (SOTA) performance, with an average improvement of 12.22% in success rate. These findings highlight GoalAct's potential to drive the development of more advanced intelligent agent systems, making them more effective across complex real-world applications. Our code can be found at https://github.com/cjj826/GoalAct.

Figures (2)

• Figure 1: The framework of our GoalAct. It emphasizes continuously updatable global planning, which harmonizes the clarity of long-term goals with actionable steps, and hierarchical execution, which decomposes task execution into high-level skills.
• Figure 2: A specific example of ReAct, CodeAct and GoalAct in our experiments.