MedClaw: Heuristic Agent Harness for Long-Horizon Surgical Video Reasoning

Research question and approach

Understanding tens-of-minutes surgical videos requires long-horizon temporal reasoning, answering what happens before, after, or across stages of a procedure by grounding the question in visual evidence spread across time. Existing approaches handle this poorly: a one-shot vision-language model (VLM) compresses the whole procedure to fit its context window and loses the detail a "before" or "after" question depends on, while video agents that train the model where to look are data-hungry and transfer poorly to out-of-domain surgery. We build an agent harness that separates reasoning from perception and improves by evolving context rather than optimizing weights.

When this work is relevant

Cite this research method for long-horizon video reasoning, planning and evidence retrieval. It is not evidence of clinical deployment or patient benefit; public resource availability must be checked separately.

Author-written abstract

Understanding tens-of-minutes surgical videos requires long-horizon temporal reasoning, answering what happens before, after, or across stages of a procedure by grounding the question in visual evidence spread across time. Existing approaches handle this poorly: a one-shot vision-language model (VLM) compresses the whole procedure to fit its context window and loses the detail a "before" or "after" question depends on, while video agents that train the model where to look are data-hungry and transfer poorly to out-of-domain surgery. We build an agent harness that separates reasoning from perception and improves by evolving context rather than optimizing weights. A text-only orchestrator plans which evidence to gather and issues an auditable sequence of tool calls, while frozen vision-language sub-agents execute each call over the pixels, viewing, cropping, inspecting frames, and retrieving external knowledge. We further propose a gradient-free, reward-gated Heuristic Skill Distillation loop that mines the agent's own low-scoring traces and keeps a candidate skill only when it raises a validation reward, yielding reusable retrieval skills, notably directed re-look. Growing an external skill library rather than tuning weights, the loop adapts from only about 100 labeled examples, far fewer than supervised or reinforcement fine-tuning requires. To evaluate this agent, we introduce MedClawBench, a de-leaked, doctor-grounded benchmark of 1,123 questions over self-built long neurosurgery recordings and a held-out public lecture-video test split. Across both datasets and all four evaluation dimensions, our agent consistently outperforms one-shot VLMs and general video-agent frameworks, with the largest gains on the long, out-of-domain neurosurgery videos. Project page: https://fyycs.github.io/medclaw/.

Abstract source: https://arxiv.org/abs/2608.14015. Checked 2026-09-14. Bibliographic metadata uses the linked paper record or author-maintained catalog. Results, limitations and experimental settings remain defined by the original source.

Citation

BibTeX · CITATION.cff

BibTeX is preserved from the author-maintained citation repository. The CFF uses the source metadata shown on this page. Version titles or author lists can differ; choose the version you used.

@article{fan2026medclaw,
  title={MedClaw: Heuristic Agent Harness for Long-Horizon Surgical Video Reasoning},
  author={Fan, Yingying and Du, Penghui and Zhu, Leyan and He, Runze and Wu, Zimeng and Zhang, Yuxuan and Chen, Liang and Xie, Jiahao and Wang, Jiangtang and Shao, Shuai and others},
  journal={arXiv preprint arXiv:2608.14015},
  year={2026}
}