NemoClaw
Run agents like Hermes, LangChain Deep Agents, and OpenClaw more securely inside NVIDIA OpenShell with managed inference
What is NemoClaw?
Run agents like Hermes, LangChain Deep Agents, and OpenClaw more securely inside NVIDIA OpenShell with managed inference
Overview of NemoClaw
What it is
NVIDIA NemoClaw is an open source reference stack for running supported AI agents more safely inside NVIDIA OpenShell sandboxes. It provides guided onboarding, managed inference, network policy, managed integrations, OpenShell-backed state persistence, and lifecycle operations through the NemoClaw CLI and its agent-specific aliases. Supported agents include OpenClaw (default), Hermes, and LangChain Deep Agents Code.
Key features
- Guided onboarding and managed inference - Network policy and egress control - OpenShell-backed state persistence - Lifecycle operations via NemoClaw CLI and agent-specific aliases - Support for OpenClaw, Hermes, and LangChain Deep Agents Code - Sandbox hardening and security best practices (documented)
- Community support and contribution guidelines (alpha project)
Use cases
- Running AI agents (OpenClaw, Hermes, LangChain Deep Agents Code) in a sandboxed environment with enhanced security - Using managed inference and network policies to control agent behavior - Developing and testing agents with OpenShell sandbox isolation and state persistence
Getting started
See the repository README for prerequisites and installation instructions. On a supported DGX or WSL host, the installer provides an express install mode that installs OpenClaw by default. For contributor setup, use the dev-setup.sh script: bash ./scripts/dev-setup.sh
License and pricing
Licensed under Apache 2.0. Open source. No pricing information is stated.
Based on the project's public README and repository metadata.