Procedural Knowledge Improves Agentic LLM Workflows

Large language models (LLMs) often struggle when performing agentictaskswithoutsubstantialtoolsupport, prom-ptengineering, or fine tuning. Despite research showing that domain-dependent, proceduralknowledgecandramaticallyincreaseplanningefficiency, little work evaluates its potential for improving LLM performance on agentic tasks that may require implicit planning. We formalize, implement, and evaluate an agentic LLM workflow that leverages procedural knowledge in the form of a hierarchical task network (HTN). Empirical results of our implementation show that handcodedHTNscandramaticallyimproveLLMperformanceonagentic tasks, and using HTNs can boost a 20b or 70b parameter LLM to outperform a much larger 120b parameter LLM baseline. Furthermore, LLM-created HTNs improve overall performance, though less so. The results suggest that leveraging expertise—from humans, documents, or LLMs—to curate procedural knowledge will become another important tool for improving LLM workflows.

Authors

• Vincent Hsiao NRC Postdoctoral Fellow, Naval Research Laboratory
• Mark Roberts Iconium Labs & NRL
• Leslie Smith Navy Center for Applied Research in AI, Naval Research Laboratory

Keywords

• Large Language Models
• Agentic Systems
• Task Networks