Skill Is Not Document: A Query-Conditional Benchmark and Two-Stage Retriever for LLM Agent Skill Routing

• URL: http://arxiv.org/abs/2606.03565v2
• Date: Mon, 08 Jun 2026 01:58:19 GMT
• Title: Skill Is Not Document: A Query-Conditional Benchmark and Two-Stage Retriever for LLM Agent Skill Routing
• Abstract summary: R3-Skill is a benchmark for realistic agent skill routing.<n>We build a two-stage retrieval system with skill compatibility as an explicit training signal.<n>The dataset, training code and model weights are released as open source for agent skill routing.
• Score: 40.648572239231804
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: LLM agents complete complex tasks by composing multiple skills, and skill retrieval is a front-end stage for agents. Skill retrieval differs fundamentally from traditional document retrieval at the supervision level: top-K joint correctness depends not only on the semantic relevance of each individual query-skill pair, but also on whether the skills retrieved together can collaborate to fulfill the task under the given query. Such "skill compatibility" cannot be derived from independent relevance alone. Yet existing LLM-based data synthesis pipelines can produce a direct supervision signal for "which skills should not be jointly retrieved under this query" -- namely the LLM's own rejection decisions -- and this signal is routinely discarded as low-quality data. To address this gap, we propose Reject-as-Resource Retriever (R3) and construct R3-Skill, a bilingual (Chinese-English) skill retrieval benchmark targeting realistic agent skill routing. R3-Skill spans four language directions, features query phrasings close to real user requests, and is verified through multi-expert cross-checking. On R3-Skill, we build a two-stage retrieval system (R3-Embedding + R3-Reranker) with skill compatibility as an explicit training signal. Gradient analysis shows that the "push-away" signal is diluted by bilateral balancing in the bi-encoder but acts as lossless graded ranking supervision in the cross-encoder -- motivating its placement at the cross-encoder stage, as confirmed by ablations on two datasets. The R3-Embedding + R3-Reranker pipeline attains Hit@1 = 0.7714, NDCG@10 = 0.8327 and Set-Compat = 0.3525 on R3-Skill. The dataset, training code and model weights are released as open source for agent skill routing.

Related papers

• Iris: Climbing to the Search Frontier [21.066541702697112]
  We present Iris-mini and Iris-pro, two search agents trained at the 35B-A3B and 397B-A17B scales, together with the data pipeline and training recipe behind them.
  arXiv  Detail & Related papers  (2026-09-03T17:51:01Z)
• Skill2Query: Exploiting Skill Structure to Generate Pseudo-Queries for Agent Skill Retrieval [23.49934184220403]
  Pseudo-Query generation can alleviate the supervision bottleneck for agent skill retrieval.<n>We propose Skill2Query, a framework that parses a skill document into a Skill Knowledge Graph and then generates pseudo-queries.<n>The generated queries can be used for offline index augmentation, online query expansion, and retriever training.
  arXiv  Detail & Related papers  (2026-08-17T04:04:19Z)
• Generative Skill Composition for LLM Agents [58.47191648555256]
  LLM agents benefit from skills for solving complex tasks.<n>As skill libraries grow, selecting an appropriate skill composition has emerged as a central bottleneck.<n>We propose SkillComposer, which instantiates structured skill composition as task-conditioned skill sequence prediction.
  arXiv  Detail & Related papers  (2026-06-30T17:53:09Z)
• Compositional Skill Routing for LLM Agents: Decompose, Retrieve, and Compose [0.0]
  We present SkillWeaver, a decompose-retrieve-compose framework combining an LLM task decomposer, a bi-encoder skill retriever with FAISS indexing, and a dependency-aware DAG planner.<n>We show that SkillWeaver reduces context window consumption by over 99%, and transfer experiments confirm generalization.
  arXiv  Detail & Related papers  (2026-06-16T15:27:55Z)
• SkillCAT: Contrastive Assessment and Topology-Aware Skill Self-Evolution for LLM Agents [42.30624901249764]
  Current pipelines learn from one trajectory per task, merge candidate skill patches before checking them, and load the full skill corpus before inference.<n>We propose SkillCAT, a training-free framework that separates this process into three stages.<n>We evaluate SkillCAT on common agent benchmarks, including SpreadsheetBench, WikiTableQuestions, and DocVQA.
  arXiv  Detail & Related papers  (2026-06-11T13:12:10Z)
• Where Do Deep-Research Agents Go Wrong? Span-Level Error Localization in Agent Trajectories [51.22051230894794]
  Evaluation based on final answers shows whether an agent succeeds, but not which parts of the trajectory make the answer unreliable.<n>We collect 2,790 real trajectories from two agent frameworks, three backbone models, and three benchmarks, convert raw logs into semantic spans, and annotate harmful error spans through expert review.<n>We propose DRIFT, a claim-centric auditing framework that tracks agent claims, checks their support in trajectory evidence, and marks spans where unsupported or conflicting claims affect the answer path.
  arXiv  Detail & Related papers  (2026-06-01T10:50:26Z)
• Cooperative Retrieval-Augmented Generation for Question Answering: Mutual Information Exchange and Ranking by Contrasting Layers [2.007262412327553]
  CoopRAG is a novel RAG framework for the question answering task in which a retriever and an LLM work cooperatively with each other.<n>Our experiments demonstrate that CoopRAG consistently outperforms state-of-the-art QA methods on three multi-hop QA datasets as well as a simple QA dataset.
  arXiv  Detail & Related papers  (2025-12-11T08:35:17Z)
• TeaRAG: A Token-Efficient Agentic Retrieval-Augmented Generation Framework [62.66056331998838]
  TeaRAG is a token-efficient agentic RAG framework capable of compressing both retrieval content and reasoning steps.<n>Our reward function evaluates the knowledge sufficiency by a knowledge matching mechanism, while penalizing excessive reasoning steps.
  arXiv  Detail & Related papers  (2025-11-07T16:08:34Z)
• R3-RAG: Learning Step-by-Step Reasoning and Retrieval for LLMs via Reinforcement Learning [60.17074283370798]
  Retrieval-Augmented Generation (RAG) integrates external knowledge with Large Language Models (LLMs) to enhance factual correctness and hallucination.<n>We propose $textbfR3-RAG$, which uses $textbfR$einforcement learning to make the LLM learn how to $textbfR$eason and $textbfR$etrieve step by step, thus retrieving comprehensive external knowledge and leading to correct answers.
  arXiv  Detail & Related papers  (2025-05-26T12:25:37Z)
• Adversarial Retriever-Ranker for dense text retrieval [51.87158529880056]
  We present Adversarial Retriever-Ranker (AR2), which consists of a dual-encoder retriever plus a cross-encoder ranker. AR2 consistently and significantly outperforms existing dense retriever methods. This includes the improvements on Natural Questions R@5 to 77.9%(+2.1%), TriviaQA R@5 to 78.2%(+1.4), and MS-MARCO MRR@10 to 39.5%(+1.3%)
  arXiv  Detail & Related papers  (2021-10-07T16:41:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.