Tool Preferences in Agentic LLMs are Unreliable
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916959581372416 |
|---|---|
| author | Faghih, Kazem Wang, Wenxiao Cheng, Yize Bharti, Siddhant Sriramanan, Gaurang Balasubramanian, Sriram Hosseini, Parsa Feizi, Soheil |
| author_facet | Faghih, Kazem Wang, Wenxiao Cheng, Yize Bharti, Siddhant Sriramanan, Gaurang Balasubramanian, Sriram Hosseini, Parsa Feizi, Soheil |
| contents | Large language models (LLMs) can now access a wide range of external tools, thanks to the Model Context Protocol (MCP). This greatly expands their abilities as various agents. However, LLMs rely entirely on the text descriptions of tools to decide which ones to use--a process that is surprisingly fragile. In this work, we expose a vulnerability in prevalent tool/function-calling protocols by investigating a series of edits to tool descriptions, some of which can drastically increase a tool's usage from LLMs when competing with alternatives. Through controlled experiments, we show that tools with properly edited descriptions receive over 10 times more usage from GPT-4.1 and Qwen2.5-7B than tools with original descriptions. We further evaluate how various edits to tool descriptions perform when competing directly with one another and how these trends generalize or differ across a broader set of 17 different models. These phenomena, while giving developers a powerful way to promote their tools, underscore the need for a more reliable foundation for agentic LLMs to select and utilize tools and resources. Our code is publicly available at https://github.com/kazemf78/llm-unreliable-tool-preferences. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_18135 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Tool Preferences in Agentic LLMs are Unreliable Faghih, Kazem Wang, Wenxiao Cheng, Yize Bharti, Siddhant Sriramanan, Gaurang Balasubramanian, Sriram Hosseini, Parsa Feizi, Soheil Artificial Intelligence Computation and Language Cryptography and Security Machine Learning Large language models (LLMs) can now access a wide range of external tools, thanks to the Model Context Protocol (MCP). This greatly expands their abilities as various agents. However, LLMs rely entirely on the text descriptions of tools to decide which ones to use--a process that is surprisingly fragile. In this work, we expose a vulnerability in prevalent tool/function-calling protocols by investigating a series of edits to tool descriptions, some of which can drastically increase a tool's usage from LLMs when competing with alternatives. Through controlled experiments, we show that tools with properly edited descriptions receive over 10 times more usage from GPT-4.1 and Qwen2.5-7B than tools with original descriptions. We further evaluate how various edits to tool descriptions perform when competing directly with one another and how these trends generalize or differ across a broader set of 17 different models. These phenomena, while giving developers a powerful way to promote their tools, underscore the need for a more reliable foundation for agentic LLMs to select and utilize tools and resources. Our code is publicly available at https://github.com/kazemf78/llm-unreliable-tool-preferences. |
| title | Tool Preferences in Agentic LLMs are Unreliable |
| topic | Artificial Intelligence Computation and Language Cryptography and Security Machine Learning |
| url | https://arxiv.org/abs/2505.18135 |