CodeAgents: A Token-Efficient Framework for Codified Multi-Agent Reasoning in LLMs

TL;DR

CodeAgents introduces a modular, token-efficient framework for multi-agent reasoning in large language models, significantly improving planning performance and scalability across diverse benchmarks.

Contribution

The paper presents a novel pseudocode-based prompting framework that codifies multi-agent reasoning, enhancing token efficiency, interpretability, and scalability in LLM-driven multi-agent systems.

Findings

Achieves 3-36% performance improvements over natural language prompts.

Sets a new state-of-the-art success rate of 56% on VirtualHome.

Reduces token usage by up to 87%, improving scalability.

Abstract

Effective prompt design is essential for improving the planning capabilities of large language model (LLM)-driven agents. However, existing structured prompting strategies are typically limited to single-agent, plan-only settings, and often evaluate performance solely based on task accuracy - overlooking critical factors such as token efficiency, modularity, and scalability in multi-agent environments. To address these limitations, we introduce CodeAgents, a prompting framework that codifies multi-agent reasoning and enables structured, token-efficient planning in multi-agent systems. In CodeAgents, all components of agent interaction - Task, Plan, Feedback, system roles, and external tool invocations - are codified into modular pseudocode enriched with control structures (e.g., loops, conditionals), boolean logic, and typed variables. This design transforms loosely connected agent plans into cohesive, interpretable, and verifiable multi-agent reasoning programs. We evaluate the proposed framework across three diverse benchmarks - GAIA, HotpotQA, and VirtualHome - using a range of representative LLMs. Results show consistent improvements in planning performance, with absolute gains of 3-36 percentage points over natural language prompting baselines. On VirtualHome, our method achieves a new state-of-the-art success rate of 56%. In addition, our approach reduces input and output token usage by 55-87% and 41-70%, respectively, underscoring the importance of token-aware evaluation metrics in the development of scalable multi-agent LLM systems. The code and resources are available at: https://anonymous.4open.science/r/CodifyingAgent-5A86