{
  "id": 1068,
  "url": "https://arxiv.org/abs/2606.14239v1",
  "title": "SkillAudit: Ground-Truth-Free Skill Evolution via Paired Trajectory Auditing",
  "summary": "Agent skills are structured procedural packages that guide frozen LLM agents in specialized workflows. Skills rarely remain sufficient after deployment: edge cases, API changes, and deployment constraints become visible only through use, making skill evolution a practical necessity. Existing methods depend on privileged feedback such as held-out validation scores, hidden test outcomes, or environment rewards -- signals often unavailable when a practitioner has only a task description and workspa",
  "authors": "Haowen Gao, Haoran Chen, Can Wang, Shasha Guo, Liang Pang, Zhaoyang Liu et al.",
  "category": "research",
  "topics": "agents-autonomy,transparency,environment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-12T08:20:09.000Z",
  "fetched_at": "2026-07-14T14:14:59.015Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/1068",
  "original_url": "https://arxiv.org/abs/2606.14239v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}