mlsanityData to 14 Oct 2026

Library Drift: Diagnosing and Fixing a Silent Failure Mode in Self-Evolving LLM Skill Libraries

arXiv 2605.19576 ↗First posted 19 May 2026cs.AI cs.CL cs.SEcited by 10

On the map: Synthetic data and self-improvement for LLMs → Self-evolving LLM agents

Abstract (arXiv)

Self-evolving skill libraries face a silent failure mode we term 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.

Key facts. Cited by 10 papers in the map (data through 14 Oct 2026). Placed in Self-evolving LLM agents, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 16 of its 22 references that are in the map. First posted on arXiv 19 May 2026.

Builds on

16 of 22 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025EvolveR: Self-Evolving LLM Agents through an Experience-Driven Lifecycle17 Oct

Other references · 15

  1. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  2. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  3. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  4. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,276
  5. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
  6. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,442
  7. 2023Voyager: An Open-Ended Embodied Agent with Large Language Modelscited by 2,191
  8. 2023Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generationcited by 1,391
  9. 2023ExpeL: LLM Agents Are Experiential Learnerscited by 876
  10. 2024Agent Workflow Memorycited by 383
  11. 2026SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Taskscited by 315
  12. 2026AutoSkill: Experience-Driven Lifelong Learning via Skill Self-Evolutioncited by 115
3 more
  1. 2025CASCADE: Cumulative Agentic Skill Creation through Autonomous Development and Evolutioncited by 34
  2. 2026Experience Compression Spectrum: Unifying Memory, Skills, and Rules in LLM Agentscited by 17
  3. 2026From Procedural Skills to Strategy Genes: Towards Experience-Driven Test-Time Evolutioncited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 10.

10 papers.

  1. 2026SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectoriescited by 11
  2. 2026The Last AI Built by Humans: Toward Genuine Recursive Self-Improvementcited by 11
  3. 2026Ratchet: How Reliable Must an LLM Judge Be to Retire a Skill?cited by 5
  4. 2026Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agentscited by 2
  5. 2026Silent Failure in LLM Agent Systems: The Entropy Principle and the Inevitable Disorder of Autonomous Agentscited by 1
  6. 2026The Blind Curator: How a Biased Judge Silently Disables Skill Retirement in Self-Evolving Agentscited by 1
  7. 2026SCOPE: Score-Isolated Agentic Optimization for Video World Modelscited by 0
  8. 2026Repo2Skill-Evo: Repository Skills Go Stale in Silencecited by 0
  9. 2026R2 Flow: Recursive Self-Improvement via Recursive Skill Evolutioncited by 0
  10. 2026EvoIn: Bridging Evolution and Internalization for Agent Fine-Tuningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Skill Drift Is Contract Violation: Proactive Maintenance for LLM Agent Skill Libraries
  2. 2026More Skills, Worse Agents? Skill Shadowing Degrades Performance When Expanding Skill Libraries
  3. 2026SkillOps: Managing LLM Agent Skill Libraries as Self-Maintaining Software Ecosystems
  4. 2026Practice Makes Unsafe: Skill Misevolution in Self-Improving LLM Agents
  5. 2026When Self-Evolution Backfires: Pre-Commit Gating against Skill Contamination in LLM Agents
  6. 2026Not All Skills Help: Measuring and Repairing Agent Knowledge
  7. 2026Rethinking Self-Evolving Agent Skills: Feedback Dynamics over Multiple Rounds
  8. 2026The Router Within: Eliciting Native Skill Routing from a Frozen LLM