One-for-All Adaptive Radiotherapy Planning Agent: A Foundation Framework for Daily CBCT-guided Radiotherapy.

Adaptive radiotherapy treatment planning seeks to enable accurate and individualized cancer care by ensuring that radiation dose conforms to the target while minimizing exposure to surrounding healthy tissues. Its success depends on accurate characterization of patient-specific anatomy at the time of treatment, whereas standard radiotherapy planning generally uses one pretreatment CT dataset for the entire treatment course, despite ongoing changes in disease extent and internal organ position. While daily adaptive radiotherapy (ART) can mitigate these discrepancies, it remains clinically impractical as each patient and fraction requires a bespoke, labor-intensive workflow. This process necessitates complex clinical decision-making-ranging from simple registration to a complete plan redesign involving CT collection, re-segmentation, and dose optimization-creating an unsustainable burden on hospital resources. In this work, we introduce the One-for-All Adaptive Radiotherapy Planning Agent, a unified foundation-model-based system that performs complete, treatment-specific online adaptive planning directly from daily cone-beam CT in under two minutes. The agent first autonomously predicts all essential planning components, including synthetic CT generation, multimodal alignment, and tumor/organ segmentation. It then intelligently leverages these outputs to execute the final clinical plan design, providing a comprehensive, automated solution for daily treatment. We also demonstrate that the agent enables clinicians to define planning with intent and intervene at critical decision points, ensuring a "human-in-the-loop" framework that generates acceptable plans before final approval. Evaluated on multiple datasets spanning head-and-neck, lung, abdominal, and prostate cancers with both photon and proton therapy, the proposed framework achieves clinically acceptable accuracy and plan quality comparable to clinically generated treatment plans, with target dose errors (