Interesting paper demystifying agent skills.

If you maintain skills for your agent, this one is worth your time.

(bookmark it)

Skills are usually assumed to inject knowledge the model lacks. However, this paper finds something interesting.

Across 8,135 normalized trial records, procedural anchoring accounts for 65.7% of cases where a skill helps, and explicit knowledge injection accounts for 4.5%. Skills stabilize execution rather than supply facts.

As the pool grows from 5 to 100 skills, actual-use precision falls from 29.6% to 3.3%.

Skills still beat Workflow Memory by 6.06 points in matched comparisons, and they break under brittle assumptions, incompatible contexts, or insufficient adaptation.

Paper: https://arxiv.org/abs/2608.14036

Track more trending AI papers in our academy: https://academy.dair.ai/