Butterfly Effects in Toolchains: A Comprehensive Analysis of Failed Parameter Filling in LLM Tool-Agent Systems

TL;DR

This paper analyzes the causes of parameter failure in LLM tool-agent systems, categorizes failure types, and proposes strategies to enhance reliability and effectiveness of tool interactions.

Contribution

It introduces a comprehensive taxonomy of parameter failures and explores their causes, providing practical suggestions for improving LLM tool-agent systems.

Findings

Parameter name hallucination mainly caused by LLM limitations

Input source issues lead to other failure patterns

Standardizing formats and improving feedback can reduce failures

Abstract

The emergence of the tool agent paradigm has broadened the capability boundaries of the Large Language Model (LLM), enabling it to complete more complex tasks. However, the effectiveness of this paradigm is limited due to the issue of parameter failure during its execution. To explore this phenomenon and propose corresponding suggestions, we first construct a parameter failure taxonomy in this paper. We derive five failure categories from the invocation chain of a mainstream tool agent. Then, we explore the correlation between three different input sources and failure categories by applying 15 input perturbation methods to the input. Experimental results show that parameter name hallucination failure primarily stems from inherent LLM limitations, while issues with input sources mainly cause other failure patterns. To improve the reliability and effectiveness of tool-agent interactions, we propose corresponding improvement suggestions, including standardizing tool return formats, improving error feedback mechanisms, and ensuring parameter consistency.