Skill∞ Lets a Meta-Agent Rewrite Its Own Skill Optimizer, Beats Human-Curated Skills on SkillsBench

Skill∞ makes agent skill optimization recursively self-improving: a meta-agent follows a meta-skill to optimize task skills, then uses the same procedure to rewrite the meta-skill itself. The meta-skill is decoupled from task skills, refined across multiple optimization problems and reused across domains. On 83 SkillsBench tasks it hits 60.7% pass rate, beating human-curated skills (56.4%) and SkillEvolver (54.5%).