github.devotd.
NewAI agents

nvidia/nemoclaw — Coding Agent Stack for the OpenShell Sandbox

nvidia/nemoclaw is a reference stack that runs coding agents such as Hermes, LangChain Deep Agents, and OpenClaw inside the NVIDIA OpenShell sandbox with managed inference.

Repositorynvidia/nemoclaw ↗

Run agents like Hermes, LangChain Deep Agents, and OpenClaw more securely inside NVIDIA OpenShell with managed inference

stars
★ 22.8K
Forks
3.1K
Language
TypeScript
License
Apache-2.0
Read original source ↗View on GitHub ↗

nvidia/nemoclaw is a reference stack that configures coding agents such as Hermes, LangChain Deep Agents, and OpenClaw to run inside NVIDIA OpenShell. It also provides managed inference, with the goal of operating agents more safely. The repository description presents OpenShell sandboxes as the execution environment, and there is no information on the latest release.

The core idea is to keep the execution and inference of multiple agents within a single configuration. What stands out is that it shows the trend of agent sandbox isolation moving onto the NVIDIA hardware stack. The repository is written in TypeScript and uses the Apache-2.0 license. It has 22756 stars. According to the README summary, Hermes is the default supported agent, and there are 373 Hermes-related files. It is also large in overall scope, covering the whole repository: it includes 3497 src files and 1847 tests.

That said, the supported agents are limited to three: Hermes, LangChain Deep Agents, and OpenClaw. Since most of the features are described on the NVIDIA documentation site, it is difficult to grasp the full usage from the repository code alone. The execution host must be a DGX or WSL, which also means it cannot be used for local development on Windows or macOS. So rather than a general-purpose agent runner, this repository is closer to an NVIDIA reference configuration with fixed supported environments and agents.