SkillFlow:Benchmarking Lifelong Skill Discovery and Evolution for Autonomous Agents
For researchers developing lifelong learning agents, this benchmark reveals that current models struggle to effectively discover and transfer skills, with high skill usage not guaranteeing utility.
SkillFlow introduces a benchmark of 166 tasks across 20 families to evaluate whether autonomous agents can discover, repair, and maintain skills over a lifetime. Experiments show a capability gap: Claude Opus 4.6 improves from 62.65% to 71.08% (+8.43 points) with lifelong skill evolution, while others like Kimi K2.5 gain only +0.60 points despite high skill usage.
As the capability frontier of autonomous agents continues to expand, they are increasingly able to complete specialized tasks through plug-and-play external skills. Yet current benchmarks mostly test whether models can use provided skills, leaving open whether they can discover skills from experience, repair them after failure, and maintain a coherent library over time. We introduce SkillFlow, a benchmark of 166 tasks across 20 families in which task construction within each family follows a Domain-Agnostic Execution Flow (DAEF) that defines an agent workflow framework, allowing these tasks to share a consistent workflow. Agents are evaluated under an Agentic Lifelong Learning protocol in which they begin without skills, solve tasks sequentially within each family, externalize lessons through trajectory- and rubric-driven skill patches, and carry the updated library forward. Experiments reveal a substantial capability gap. For Claude Opus 4.6, lifelong skill evolution improves task success from 62.65% to 71.08% (+8.43 points). However, high skill usage does not necessarily imply high utility: Kimi K2.5 gains only +0.60 points despite 66.87% skill usage, while Qwen-Coder-Next reaches only a 44.58% task completion rate and still regresses relative to the vanilla setting. SkillFlow contributes a structured testbed for this direction and an in-depth empirical analysis of skill discovery, patching, transfer, and their failure modes under lifelong evaluation.