mlsanityData to 7 Oct 2026

Agent-Facing Information Design in LLM Tool Registries

arXiv 2605.23916 ↗First posted 12 Apr 2026cs.IR cs.AI econ.GNcited by 0

On the map: LLM agents and tool use → Financial trading and forecasting agents

Abstract (arXiv)

AI agents often pick tools from registries, where each tool's provider writes its description. We ask whether sales language in those descriptions changes which tool an agent picks. We built pairs of listings differing in one controlled way (added praise, a verifiable specification, or list order) and asked two OpenAI models to call one tool. In a preregistered study, stacked praise (four kinds combined) raised a tool's pick rate by about 43 percentage points, matching or beating a verifiable specification. Praise also pulled some picks toward tools that could not do the task, but rarely toward tools asking for unneeded data access. With identical listings, the first-listed tool was picked about 72 points more often. On tasks with numeric limits, structured fields helped agents pick the capable tool; adding the provider's sales text beside the fields reduced or erased that gain. Registries could list limits as fields, hide sales text from agents, and randomize order. Stacked praise, but no single kind, replicated on held-out domains. Results are provisional until blind phrase ratings are complete, and cover two small models.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Financial trading and forecasting agents, one of 16 directions in LLM agents and tool use. Builds on 15 of its 25 references that are in the map. First posted on arXiv 12 Apr 2026.

Builds on

15 of 25 references.

References · 15

  1. 2025Prompt Injection Attack to Tool Selection in LLM Agentscited by 145
  2. 2024Adversarial Search Engine Optimization for Large Language Modelscited by 37
  3. 2024From Allies to Adversaries: Manipulating LLM Tool-Calling through Adversarial Injectioncited by 36
  4. 2025MPMA: Preference Manipulation Attack Against Model Context Protocolcited by 31
  5. 2024Manipulating Large Language Models to Increase Product Visibilitycited by 24
  6. 2025Attractive Metadata Attack: Inducing LLM Agents to Invoke Malicious Toolscited by 23
  7. 2025ToolTweak: An Attack on Tool Selection in LLM-based Agentscited by 22
  8. 2025What Is Your AI Agent Buying? Evaluation, Biases, Model Dependence, & Emerging Implications for Agentic E-Commercecited by 18
  9. 2026Model Context Protocol (MCP) Tool Descriptions Are Smelly! Towards Improving AI Agent Efficiency with Augmented MCP Tool Descriptionscited by 15
  10. 2025Tool Preferences in Agentic LLMs are Unreliablecited by 13
  11. 2025MCPGuard : Automatically Detecting Vulnerabilities in MCP Serverscited by 13
  12. 2025Bias Beware: The Impact of Cognitive Biases on LLM-Driven Product Recommendationscited by 10
3 more
  1. 2025A Framework for Studying AI Agent Behavior: Evidence from Consumer Choice Experimentscited by 10
  2. 2025BiasBusters: Uncovering and Mitigating Tool Selection Bias in Large Language Modelscited by 8
  3. 2025Dynamics of Adversarial Attacks on Large Language Model-Based Search Enginescited by 5

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025InfoBid: A Simulation Framework for Studying Information Disclosure in Auctions with Large Language Model-based Agents
  2. 2026LLMAdBench: A Human Preference Benchmark for Advertising in LLM Responses
  3. 2026Mechanism Design for Quality-Preserving LLM Advertising
  4. 2026AgentFairBench: Do LLM Agents Discriminate When They Act?
  5. 2026Paying for Honesty Without Knowing the Truth: Reputation-Penalty Design for LLM Marketplace Agents
  6. 2026Whom Do AI Agents Work For? Role Assignment Induces Sponsorship Bias in LLM Recommenders
  7. 2026Auditing Provenance Sensitivity in LLM Agent Action Selection
  8. 2026Commercial Persuasion in AI-Mediated Conversations