{
  "id": 5706,
  "url": "https://arxiv.org/abs/2604.16753v1",
  "title": "Know When to Trust the Skill: Delayed Appraisal and Epistemic Vigilance for Single-Agent LLMs",
  "summary": "As large language models (LLMs) transition into autonomous agents integrated with extensive tool ecosystems, traditional routing heuristics increasingly succumb to context pollution and \"overthinking\". We argue that the bottleneck is not a deficit in algorithmic capability or skill diversity, but the absence of disciplined second-order metacognitive governance. In this paper, our scientific contribution focuses on the computational translation of human cognitive control - specifically, delayed a",
  "authors": "Eren Unlu",
  "category": "research",
  "topics": "regulation,agents-autonomy",
  "orgs": null,
  "regions": null,
  "published_at": "2026-04-17T23:55:19.000Z",
  "fetched_at": "2026-07-14T16:31:57.534Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/5706",
  "original_url": "https://arxiv.org/abs/2604.16753v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}