Genesis
SkillCloud & infraUse when working on Genesis simulation, RL teacher training, visual demo generation, or related serverless/EGL behavior.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Genesis skill
What this skill tells your AI
The instructions your AI receives, as published by nebius/nebius-physical-ai in skills/tools/genesis/SKILL.md and read by ahel’s review.
Genesis is the GPU-accelerated physics simulation tool. It uses vendored genesis.ext.pyrender plus PyOpenGL EGL.
Current Capability
- RL teacher training works headless on Nebius AI Jobs with Mesa llvmpipe fallback.
- Serverless RL teacher training works.
- 480x640 targeted renders pass via Mesa; scale performance is untested.
Rendering Gap
Visual demo generation is blocked on EGL/DRI device access in containers. libEGL_nvidia.so.0 is absent despite graphics capability environment variables, so workloads fall back to Mesa, which cannot scale.
Serverless visual demo generation is blocked by the same EGL/DRI issue.
BatchRenderer/Madrona is deferred; do not prioritize it. There is no platform escalation pending and no BUNDLE_EGL_NVIDIA action item.
Validation
E2E: 8/8 passing except visual demo generation.
Signals
- GitHub stars
- 28
- Forks
- 15
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
genesis-nebius- Source
- github.com/nebius/nebius-physical-ai