自进化模型技能库会因失控积累导致性能下降，本文提出诊断与修复方案。
Library Drift: Diagnosing and Fixing a Silent Failure Mode in Self-Evolving LLM Skill Libraries
- 通过消融实验定位技能漂移的触发机制
- 修复后代码任务通过率从25.8%提升至58.4%
- 适合自进化AI系统开发者参考
自进化技能库存在一种隐蔽的失败模式——‘技能漂移’：在无结果驱动的生命周期管理下，技能无限累积导致检索退化、误报注入和性能停滞。近期评估显示，由大模型生成的技能仅带来0.0个百分点提升，而人工精调技能则带来16.2个百分点提升（SkillsBench）。本文提出三步方案：(1) 可复现的触发机制：消融实验证明，禁用技能注入（基准+0.002）或过早淘汰技能（造成-0.019伤害）均会引发漂移；(2) 追踪级诊断工具：基于追加式证据日志，包含每项技能贡献分、归因判断及路由参与度，可在最终任务得分前暴露问题；(3) 经验证的修复方案：采用结果驱动的淘汰策略 + 有限活跃容量 + 元技能创作前置，使MBPP+ hard-100任务在100轮中持留通过率从0.258提升至0.584（滚动提升+0.328）。八组消融分析揭示各治理机制的承重作用，提供可复用的诊断手册。
原文摘要 · Abstract (English)
Self-evolving skill libraries face a silent failure mode we term \emph{library drift}: unbounded skill accumulation without outcome-driven lifecycle management causes retrieval degradation, false-positive injections, and performance stagnation. Recent evaluation confirms the symptom (LLM-authored skills deliver +0.0pp gain while human-curated ones deliver +16.2pp (SkillsBench)), yet the underlying mechanism has not been isolated. We provide (1) a reproducible trigger: ablations that isolate drift: one disables skill injection (flat floor, +0.002), one imposes premature retirement (active harm, $-$0.019); (2) trace-level diagnostics: an append-only evidence log with per-skill contribution scores, attribution verdicts, and router engagement metrics that make the failure visible before it reaches end-task scores; and (3) a verified fix: a minimal governance recipe (outcome-driven retirement + bounded active-cap + meta-skill authoring prior) that lifts held-out pass@1 from a 0.258 baseline to a late-window mean of 0.584 (rolling gain $+$0.328) on MBPP+ hard-100 over 100 rounds. Eight ablations decompose which governance mechanisms are load-bearing and which are subsumed, providing a concrete playbook for diagnosing library drift in any self-evolving agent.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。