Tools Are Not Islands: Set-Level Tool Retrieval for LLM Agents via Query-Conditioned Hyperedge Prediction
Source: https://arxiv.org/abs/2607.25718
arXiv: 2607.25718
Authors: Xinyi Hong, Pinjun Dong, Xinyang Yu, Binyan Jiang
Published: 2026-07-28
Categories: cs.LG, cs.AI, cs.IR
PDF: https://arxiv.org/pdf/2607.25718v1
Abstract (en)
Large language model (LLM) agents increasingly rely on invoking external tools to complete real-world tasks. Tool retrieval, which selects a small task-relevant subset from a library of thousands of tools before the agent acts, has therefore become a critical component of LLM agent pipelines. However, existing retrievers either score each tool in isolation or assemble the tool set sequentially, so the joint utility of a candidate set is never evaluated as a whole. In this paper, we propose HYSET, short for HYperedge-based SEt-level Tool retrieval. Our contributions are threefold: (i) we formulate tool retrieval as query-conditioned hyperedge prediction on a tool co-invocation hypergraph, under which the tool set itself becomes the unit of scoring and most existing retrieval paradigms reduce to restricted instances; (ii) we capture size-dependent tool compatibility through cardinality-specific interactions; and (iii) we design HYSET as a pre-selection module requiring no modification to the downstream agent. Experiments on ToolBench demonstrate that HYSET consistently outperforms state-of-the-art baselines in both tool retrieval performance and end-to-end task success. Beyond the in-domain setting, HYSET further supports zero-shot/few-shot transfer, generalizing to held-out tools/categories and unseen domains with minimal supervision.
Summary (zh)
HYSET 指出真实 Agent 会从上千工具中预选一小撮，但现有检索要么逐工具打分要么顺序拼集合，从不评估候选集合的联合效用。作者把 tool retrieval 形式化为 tool co-invocation 超图上的 query-conditioned hyperedge prediction，并以基数相关交互刻画不同集合大小下的兼容性；作为 pre-selection 模块无需改下游 Agent。ToolBench 上检索与端到端成功率均优于 SOTA，并支持 held-out tools/categories 与跨域 zero/few-shot。
One-liner
工具检索的打分单位应是集合：HYSET 用查询条件下的超边预测优化联合效用，而不是单工具 top-k。
Obsidian evidence
OpenClaw定时任务/论文流水线/2026-07-29-论文流水线.md- opencli:
arxiv paper 2607.25718 -f json - Run date: 2026-07-29