Search papers, labs, and topics across Lattice.
This paper introduces Capability Pages, a novel approach to skill retrieval in large language model agents that addresses the limitations of traditional retrieval methods by formalizing a skill's capability as its executable region. By employing cluster-contrastive representations that include positive triggers, negative boundaries, and a discriminative body, the authors enhance the retrieval process to better differentiate between similar skills. The proposed method demonstrates significant improvements in recall and task success across multiple datasets, achieving an average gain of 2.94 points in Recall@10 and 3.62 points in end-to-end task success.
Retrieving skills from a crowded library just got smarter鈥擟apability Pages boost retrieval accuracy by distinguishing between confusable alternatives.
As skill libraries grow, large language model agents must retrieve reusable skills from candidates that often share the same topic and vocabulary but implement different capabilities. Retrieval is limited not only by the scorer but also by the text being scored: a document may describe what a skill does without stating which similar requests should be routed elsewhere. We formalize a skill's capability as its \emph{executable region}, the set of queries it can solve, and view its document as a lossy observation of that region. This view exposes a document-imposed component of retrieval error that cannot be removed by improving the retriever alone. We therefore propose \emph{Capability Pages}, cluster-contrastive skill representations containing a positive trigger $\Tpos$, a negative boundary $\Tneg$, and a discriminative body $B$. An offline compiler compares neighboring skills to write these fields. At inference time, the index uses $\Tpos$ and $B$ for candidate recall, while the router uses $\Tneg$ to reject confusable alternatives. On SRA-Bench, which contains 26{,}262 skills and 5{,}400 questions from six datasets, Capability Pages improve Recall@10 for all five tested retrievers, with a mean gain of $2.94$ points. Adding $\Tneg$ to candidate cards improves end-to-end task success by $3.62$ points on average across four executors and six datasets. A transfer evaluation on Chinese SSL-SkillDiscovery reaches $73.07\%$ MRR@50 using the same encoder across conditions. Capability Pages require no modification to the online models; they improve routing by rewriting the offline skill library.
