AgentSpec: Understanding Embodied Agent Scaffolds Through Controlled Composition

LLM agents are increasingly built not as single model calls, but as scaffolded systems that combine reasoning, memory, reflection, action execution, and learning. While such scaffolds often improve performance, they are often embedded in tightly coupled pipelines, making it difficult to isolate component contributions, compare alternative designs, or understand how module interactions shape agent behavior. We introduce AgentSpec, a modular specification framework that represents embodied agents as typed compositions of reusable policy components with standardized interfaces. AgentSpec standardizes the interfaces among perception, memory, reasoning, reflection, action, and optional learning, enabling components to be swapped and recombined under controlled conditions. We instantiate this framework across DeliveryBench, ALFRED, MiniGrid, and RoboTHOR, and analyze reasoning, memory, reflection, and reinforcement-learning modules across model backbones. Our results show that agent performance is governed by scaffold compatibility and interaction effects rather than isolated module strength. In particular, structured multi-granularity memory improves long-horizon state tracking, reasoning and memory interact non-uniformly across environments, reflection trades off correction and cost, and RL-trained policies compose best when optimized with deployment-time scaffold structure. AgentSpec provides a controlled foundation for studying, comparing, and designing composable LLM agents. Our code, baselines and interactive playground are publicly available at https://agentspec-embodied.github.io.

Are Vision xLSTM Embedded UNet More Reliable in Medical 3D Image Segmentation?

The advancement of developing efficient medical image segmentation has evolved from initial dependence on Convolutional Neural Networks (CNNs) to the present investigation of hybrid models that combine CNNs with Vision Transformers. Furthermore, there is an increasing focus on creating architectures that are both high-performing in medical image segmentation tasks and computationally efficient to be deployed on systems with limited resources. Although transformers have several advantages like capturing global dependencies in the input data, they face challenges such as high computational and memory complexity. This paper investigates the integration of CNNs and Vision Extended Long Short-Term Memory (Vision-xLSTM) models by introducing a novel approach called UVixLSTM. The Vision-xLSTM blocks captures temporal and global relationships within the patches extracted from the CNN feature maps. The convolutional feature reconstruction path upsamples the output volume from the Vision-xLSTM blocks to produce the segmentation output. Our primary objective is to propose that Vision-xLSTM forms a reliable backbone for medical image segmentation tasks, offering excellent segmentation performance and reduced computational complexity. UVixLSTM exhibits superior performance compared to state-of-the-art networks on the publicly-available Synapse dataset. Code is available at: https://github.com/duttapallabi2907/UVixLSTM