Official code base for LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels

Research on Coding Agents

The fastest repo in history to surpass 50K stars ⭐, reaching the milestone in just 2 hours after publication. Better Harness Tools that make real things done. Now writing in Rust using oh-my-codex.

Source Code for MIRROR: Visual Motion Imitation via Real-time Retargeting and Teleoperation on Humanoid Robots with Parallel Differential Inverse Kinematics

[CoRL 2025] TWIST: Teleoperated Whole-Body Imitation System

Code for "GVHMR: World-Grounded Human Motion Recovery via Gravity-View Coordinates", Siggraph Asia 2024

[ICRA 2026] GMR: General Motion Retargeting. Retarget human motions into diverse humanoid robots in real time on CPU. Retargeter for TWIST.

Windows kernel-mode driver emulating well-known USB game controllers.

D2E: Scaling Vision-Action Pretraining on Desktop Data for Transfer to Embodied AI [ICLR 2026]

[CVPR 2026] Real2Edit2Real: Generating Robotic Demonstrations via a 3D Control Interface

Solve Visual Understanding with Reinforced VLMs

StableWorld: Towards Stable and Consistent Long Interactive Video Generation

Official implementation of Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data

RoboClaw is an Embodied AI Assistant.

A Vision-Language Model for Spatial Affordance Prediction in Robotics

[CVPR 2025] RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete. Official Repository.

Official implementation of OpenTrack.

Code for kai0, including training, inference and data collection.

We release Evo-RL, the opensource real-world offline RL on So-101 and AgileX PiPER for easier reproduction.

Whole Body Controller & Finite State Machine

The awesome collection of OpenClaw skills. 5,400+ skills filtered and categorized from the official OpenClaw Skills Registry.🦞

A set of ready to use Agent Skills for research, science, engineering, analysis, finance and writing.

Claude Code+Obsidian，邪修读论文就是快

MOVA: Towards Scalable and Synchronized Video–Audio Generation

[ICLR2026] Any-to-Bokeh is a novel one-step video bokeh framework that converts arbitrary input videos into temporally coherent, depth-aware bokeh effects.

[ICCV 2025] Decouple and Track: Benchmarking and Improving Video Diffusion Transformers for Motion Transfer