OpenShell
Kernel-level sandbox that decides what agents may touch
Security scanner for third-party AI skills
A security analysis tool from NVIDIA for inspecting AI skills before an agent uses them. It is aimed at developers and teams who install skill bundles from the internet and want to catch malicious or risky behavior first.
You supply a skill bundle, and the tool runs multiple security scanners across its files and behavior. It looks for risks such as data exfiltration, hidden instructions, or other suspicious activity, then reports what it finds so you can decide whether to trust or recreate the skill.
Free ai video creation + OCR full books + private photos + more Github finds
Free CapCut, marketing tools, and all of github’s top 10 repos
Kernel-level sandbox that decides what agents may touch
Persistent Linux sandbox for safe agentic development
Fast hardware-isolated VMs for untrusted AI code
Small composable agent skills that give engineers better habits
A whole power user's Claude Code setup in one install
Spec-driven development toolkit that makes agents agree before coding