{
  "schema_version": "1.1",
  "id": "archive:https://arxiv.org/abs/2608.16889v1",
  "slug": "2608-16889v1-17rof2d",
  "url": "https://feed7.dev/p/2608-16889v1-17rof2d",
  "title": "Don't Drop the BATON: Long-Horizon Robot Manipulation via Agentic Subtask Exploration and Transition-aware Memory",
  "why_included": "BATON explores robot subtasks independently, stores their solutions, and manages handoffs between them, replacing multiplicative whole-task search with additive composition.",
  "summary": "BATON freezes the vision-language-action model and explores each subtask separately, changing estimated exploration from **T^K to T×K**. Transition-aware memory checks invocation readiness, repairs handoff state, and selects outcomes compatible with the next stage without updating parameters.",
  "practical_implication": "The transferable agent pattern is to make subtasks independently testable and reusable, then represent their entry and exit contracts explicitly. That localizes failures and prevents a step that passes in isolation from leaving unusable state downstream.",
  "agent_context": "BATON freezes the vision-language-action model and explores each subtask separately, changing estimated exploration from **T^K to T×K**. Transition-aware memory checks invocation readiness, repairs handoff state, and selects outcomes compatible with the next stage without updating parameters.\n\nThe transferable agent pattern is to make subtasks independently testable and reusable, then represent their entry and exit contracts explicitly. That localizes failures and prevents a step that passes in isolation from leaving unusable state downstream.\n\nOn RoboMemArena, BATON reports **+11.6% task success** and **+14.9% cumulative success** over the prior state of the art. The evidence is specific to long-horizon robot manipulation, so applicability to software-agent workflows remains an analogy rather than a measured result.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2608.16889v1",
    "published_at": "2026-08-17T17:59:57.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "agent",
  "domains": [],
  "topics": [
    "subagents",
    "agent-memory",
    "harness-engineering"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "On RoboMemArena, BATON reports **+11.6% task success** and **+14.9% cumulative success** over the prior state of the art. The evidence is specific to long-horizon robot manipulation, so applicability to software-agent workflows remains an analogy rather than a measured result."
  ],
  "connected_context": {
    "meaning": "BATON gives the prior harness guidance a more precise implementation pattern: independently exercise bounded subtasks, but treat each handoff as a checked state contract rather than assuming local success composes. Its robotics results support that architecture within one embodied domain; they do not establish the same gains for software or multi-agent workflows.",
    "corpus_size": 479,
    "generated_at": "2026-08-18T10:05:01.713Z",
    "connections": [
      {
        "title": "Anthropic's CCA Exam as a Field-Guide for Agentic Engineering — Frank Coyle, UC Berkeley",
        "source_name": "AI Engineer",
        "source_url": "https://www.youtube.com/watch?v=Z-c11pV_uvU",
        "feed7_url": "https://feed7.dev/p/anthropic-s-cca-exam-as-a-field-guide-for-agentic-engineering-frank-coyl-00nf896",
        "reason": "BATON operationalizes the field guide’s bounded-subagent and output-isolation advice by checking whether each subtask’s resulting state is actually usable by the next."
      },
      {
        "title": "Why We Killed Our Multi-Agent Pipeline — Subbiah Sethuraman and Abhilash Asokan, ZS Associates",
        "source_name": "AI Engineer",
        "source_url": "https://www.youtube.com/watch?v=u6jJcIFDLE4",
        "feed7_url": "https://feed7.dev/p/why-we-killed-our-multi-agent-pipeline-subbiah-sethuraman-and-abhilash-a-0fmz3z3",
        "reason": "The failed fixed pipeline identifies context loss and incoherent handoffs; BATON addresses the analogous composition failure with explicit transition readiness and outcome selection."
      },
      {
        "title": "A Taxonomy of Cognitive Capability Gaps in Generative and Agentic AI",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2608.02553v1",
        "feed7_url": "https://feed7.dev/p/2608-02553v1-12y8joy",
        "reason": "BATON supplies a concrete, empirically tested robotics mechanism for parts of the taxonomy’s durable-state, bounded-goal, monitoring, and adaptation requirements."
      }
    ]
  },
  "lifecycle": "Current",
  "published_at": "2026-08-17T17:59:57.000Z",
  "modified_at": "2026-08-17T17:59:57.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2608-16889v1-17rof2d",
    "json": "https://feed7.dev/p/2608-16889v1-17rof2d.json",
    "markdown": "https://feed7.dev/p/2608-16889v1-17rof2d.md"
  }
}