40,285 Skills Later, Supply Still Doesn't Match Demand

Notes on Agent Skills: A Data-Driven Analysis of Claude Skills for Extending Large Language Model Functionality (arXiv:2602.08004) — G. Ling, Shan Zhong, R. Huang · February 2026

Note published July 25, 2026 · written by SkillFed’s research pipeline from the paper above · how these notes are made

Ling, Zhong, and Huang treat a live marketplace as a dataset, not a sample. All 40,285 publicly listed agent skill listings get pulled and measured — when they're published, what category they land in, how long they run, how much they're actually used, how similar they are to each other, and what actions they're allowed to take. The payoff is a corpus-scale baseline for what an agent skill ecosystem looks like in practice, not aspiration.

New skills don't arrive on a steady drip. Publication comes in bursts, each one riding a wave of attention around some topic before it fades. Software engineering dominates raw supply, but information retrieval and content-creation skills pull in adoption far out of proportion to their numbers — the paper's headline finding, a category-level supply-demand imbalance. Skill length is heavy-tailed, though the typical skill still fits inside normal prompt budgets without trouble. Underneath the catalog's size sits real overlap, too: heavy intent-level redundancy in what skills are actually trying to do, plus a meaningful slice of listings that grant state-changing or system-level actions — risk sitting right next to skills that only read and summarize.