Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-02-17 · source-backed
SkillsBench is the first standardized benchmark for evaluating agent skills — 86 tasks across 11 domains, 7,308 test trajectories. The critical finding: curated skills boost agent pass rates by 16.2%, but self-generated skills provide zero benefit on average. Smaller models with good skills can match larger models without them. This validates investing time in hand-crafted skills over auto-generated ones and scientifically justifies the value of skill marketplaces.
Each link below shares sources, entities, or timing with this story.
Same source / Shared topic / Tension
Cite the same source (SkillsBench); overlapping topics (agent, benchmark, critical, curated, finding); pushes against this story (against).
Shared entity: Zero / Same source domain / Shared topic / What happened next
Both cover Zero; reported by the same outlet (github.com); overlapping topics (agent, skill, zero).
Shared entity: SkillsBench / Shared topic / What happened next / Tension
Both cover SkillsBench; overlapping topics (agent, benchmark, model); picks up the SkillsBench thread on 2026-04-21.
Shared entity: Zero / Shared topic / What happened next / Tension
Both cover Zero; overlapping topics (model, skill); picks up the Zero thread on 2026-05-10.
Both cover Zero; overlapping topics (agent, ai-generated); picks up the Zero thread on 2026-03-08.
Same source domain / Shared topic / Tension
Reported by the same outlet (github.com); overlapping topics (agent, critical, finding, skill); pushes against this story (but).
Shared entity: SkillsBench / Same source
Both cover SkillsBench; cite the same source (SkillsBench).
Shared entity: Zero / Same source domain / Shared topic
Both cover Zero; reported by the same outlet (github.com); overlapping topics (agent, skill).