Saved in:
Bibliographic Details
Main Authors: Zheng, Wei, Yan, Yang, Shao, Yiyang, Li, Jinyang, Chang, Zeze, Jia, Yukuang, Mao, Qiming, Wang, Chihyung, Zhou, Jingbin
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.29270
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913169619812352
author Zheng, Wei
Yan, Yang
Shao, Yiyang
Li, Jinyang
Chang, Zeze
Jia, Yukuang
Mao, Qiming
Wang, Chihyung
Zhou, Jingbin
author_facet Zheng, Wei
Yan, Yang
Shao, Yiyang
Li, Jinyang
Chang, Zeze
Jia, Yukuang
Mao, Qiming
Wang, Chihyung
Zhou, Jingbin
contents The era of the Internet of Agents (IoA) is taking shape: LLM agents are expected to fulfill user goals by orchestrating fast-growing populations of Model Context Protocol (MCP) servers, Agent-to-Agent (A2A) endpoints, reusable skills, and other LLM-callable services. Yet LLMs face a structural mismatch with this regime: effective context is a scarce resource that does not scale with the number of services. Concatenating thousands of service descriptions into a prompt overflows the context window, and even when the window is large enough, models systematically under-attend to information in the middle of long inputs, the well-documented Lost-in-the-Middle phenomenon. This is fundamentally a question of context management for service discovery. To address this, we propose an LLM-native progressive-disclosure scheme and its concrete instantiation, A2X (Agent-to-Anything service discovery): an LLM-driven pipeline that automatically organizes the registered services into a hierarchical taxonomy and walks it layer by layer at query time, so that every LLM call sees only a small candidate set highly relevant to the user query. This decouples effective-context scarcity from registry size and significantly reduces token consumption while improving retrieval accuracy. Compared to full-context dumping, A2X achieves a 6.2-point Hit Rate gain at one-ninth the prompt-token cost; compared to the state-of-the-art open-source embedding-based baseline, A2X improves Hit Rate by more than 20 points.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29270
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Indexing the Unreadable: LLM-Native Recursive Construction and Search of Service Taxonomies
Zheng, Wei
Yan, Yang
Shao, Yiyang
Li, Jinyang
Chang, Zeze
Jia, Yukuang
Mao, Qiming
Wang, Chihyung
Zhou, Jingbin
Artificial Intelligence
The era of the Internet of Agents (IoA) is taking shape: LLM agents are expected to fulfill user goals by orchestrating fast-growing populations of Model Context Protocol (MCP) servers, Agent-to-Agent (A2A) endpoints, reusable skills, and other LLM-callable services. Yet LLMs face a structural mismatch with this regime: effective context is a scarce resource that does not scale with the number of services. Concatenating thousands of service descriptions into a prompt overflows the context window, and even when the window is large enough, models systematically under-attend to information in the middle of long inputs, the well-documented Lost-in-the-Middle phenomenon. This is fundamentally a question of context management for service discovery. To address this, we propose an LLM-native progressive-disclosure scheme and its concrete instantiation, A2X (Agent-to-Anything service discovery): an LLM-driven pipeline that automatically organizes the registered services into a hierarchical taxonomy and walks it layer by layer at query time, so that every LLM call sees only a small candidate set highly relevant to the user query. This decouples effective-context scarcity from registry size and significantly reduces token consumption while improving retrieval accuracy. Compared to full-context dumping, A2X achieves a 6.2-point Hit Rate gain at one-ninth the prompt-token cost; compared to the state-of-the-art open-source embedding-based baseline, A2X improves Hit Rate by more than 20 points.
title Indexing the Unreadable: LLM-Native Recursive Construction and Search of Service Taxonomies
topic Artificial Intelligence
url https://arxiv.org/abs/2605.29270