Skills for Real Engineers. Straight from my .agents directory.

Official Codebase for "Do as I Do: Dexterous Manipulation Data from Everyday Human Videos"

A Web-Scale 4D Hand-Object Interaction Data Engine for Any-View Robot Retargeting and Video-to-Action Robot Learning

Provide with pre-build flash-attention 2 and 3 package wheels on Linux and Windows using GitHub Actions

A curated, continuously updated reading list, paper blogs, and resources for World Action Models (WAMs) in embodied AI.

A paper list for Learning-based 3D Vision.

A practical toolkit for process-level robot evaluation with Process Reward Models (PRMs).

A paper list of Awesome Latent Space.

📚 A curated collection of papers and open-source code repositories dedicated to the application of Vision-Language Models (VLMs) for streaming video.

Panoramic Affordance Prediction (PAP) (ECCV 2026)

[RSS 2026] Interactive World Simulator for Robot Policy Training and Evaluation

[CVPR 2025] VideoWorld is a simple generative model that learns purely from unlabeled videos—much like how babies learn by observing their environment.

Code to pretrain, fine-tune, and evaluate DreamZero and run sim & real-world evals

[NeurIPS 2025 Spotlight] Towards Understanding Camera Motions in Any Video

CamReasoner: Reinforcing Camera Movement Understanding via Structured Spatial Reasoning

[RSS 2026] Causal video-action world model for generalist robot control

Cosmos Policy

Being-H is BeingBeyond's family of human-centric embodied foundation models.

SPAgent, a foundation agent for understanding, reasoning over, and operating within the physical and spatial world.

Orient Anything V2, NeurIPS 2025 Spotlight

[ICCV 2025] VLM4D: Towards Spatiotemporal Awareness in Vision Language Models

Official code for "TraceGen: World Modeling in 3D Trace-Space Enables Learning from Cross-Embodiment Videos" (CVPR 2026)

[NeurIPS 2025] Pixel-Perfect Depth

PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulation