{
  "schema_version": "1.1",
  "id": "archive:https://arxiv.org/abs/2608.20274v1",
  "slug": "2608-20274v1-08ujug8",
  "url": "https://feed7.dev/p/2608-20274v1-08ujug8",
  "title": "Break It Down, Pass It On: Cross-Task Skill Transfer in LLM Agents",
  "why_included": "Reusable agent skills transferred better when induced from subtasks and stored as text. Whole-task skills often hurt versus no memory, so retrieval alone is not evidence that a memory helps.",
  "summary": "The study compares task-level against subtask-level skill induction and text against code formats. **Task-level skills often fell below the no-memory baseline**, while **subtask-level skills improved performance on average**; text transferred better than code.",
  "practical_implication": "Build agent memory from small, reusable procedures and prefer textual guidance unless execution requires code. Before retrieval, score candidates for both task specificity and breadth of relevance rather than assuming any prior solution will help.",
  "agent_context": "The study compares task-level against subtask-level skill induction and text against code formats. **Task-level skills often fell below the no-memory baseline**, while **subtask-level skills improved performance on average**; text transferred better than code.\n\nBuild agent memory from small, reusable procedures and prefer textual guidance unless execution requires code. Before retrieval, score candidates for both task specificity and breadth of relevance rather than assuming any prior solution will help.\n\nNeither specificity nor abstractness predicted outcomes alone. Their combined **skill utility score** correlated with transfer performance and requires only skills plus task descriptions, but the supplied material gives no tasks, models, or effect sizes for judging generality.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2608.20274v1",
    "published_at": "2026-08-20T17:12:08.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "agent",
  "domains": [],
  "topics": [
    "skills",
    "agent-memory",
    "agent-reliability"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "Neither specificity nor abstractness predicted outcomes alone. Their combined **skill utility score** correlated with transfer performance and requires only skills plus task descriptions, but the supplied material gives no tasks, models, or effect sizes for judging generality."
  ],
  "connected_context": {
    "meaning": "This makes accumulated agent learning more selective than the prior “capture every lesson” pattern: reusable memory should be decomposed into small textual procedures and retrieved by joint specificity and breadth. It also gives a structural explanation for skill regressions—whole-task solutions can transfer worse than no memory—while leaving the proposed utility score’s generality unresolved without task and effect-size details.",
    "corpus_size": 525,
    "generated_at": "2026-08-22T21:13:03.535Z",
    "connections": [
      {
        "title": "The Regression Tax: Decomposing Why Skills Help and Hurt LLM Agents",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2607.22520v1",
        "feed7_url": "https://feed7.dev/p/2607-22520v1-0mz9wnf",
        "reason": "It reinforces the documented regression tax and narrows a likely cause: task-level skills often hurt, whereas smaller subtask procedures transfer better on average."
      },
      {
        "title": "SkillProx: Self-Evolving Agent Skills via Proximal Textual Gradient Descent",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2608.07449v1",
        "feed7_url": "https://feed7.dev/p/2608-07449v1-1co8km3",
        "reason": "SkillProx supplies the maintenance consequence of this result: candidate procedures should be tested for transfer, rolled back on regression, and pruned when utility is low."
      },
      {
        "title": "The Era of Compound Engineering — Kieran Klaassen, Every/Cora",
        "source_name": "AI Engineer",
        "source_url": "https://www.youtube.com/watch?v=_ehJyfHg1Vk",
        "feed7_url": "https://feed7.dev/p/the-era-of-compound-engineering-kieran-klaassen-every-cora-053chml",
        "reason": "It qualifies compound engineering’s lesson-capture strategy by showing that persistence alone is insufficient; lessons need subtask-level decomposition and relevance-aware retrieval."
      },
      {
        "title": "EveryInc/compound-engineering-plugin",
        "source_name": "GitHub",
        "source_url": "https://github.com/EveryInc/compound-engineering-plugin",
        "feed7_url": "https://feed7.dev/p/compound-engineering-plugin-06exf60",
        "reason": "The plugin is a concrete repository workflow for durable skills, while this study suggests its captured plans and solved-problem notes should be distilled into small textual procedures rather than reused wholesale."
      }
    ]
  },
  "lifecycle": "Current",
  "published_at": "2026-08-20T17:12:08.000Z",
  "modified_at": "2026-08-20T17:12:08.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2608-20274v1-08ujug8",
    "json": "https://feed7.dev/p/2608-20274v1-08ujug8.json",
    "markdown": "https://feed7.dev/p/2608-20274v1-08ujug8.md"
  }
}