For AI agents: a documentation index is available at the root level at /llms.txt and /llms-full.txt. Append /llms.txt to any URL for a page-level index, or .md for the markdown version of any page.
User Guide
User Guide
    • Home
      • Overview
      • Architecture Overview
      • Ecosystem
      • Release Notes
      • Prerequisites
      • Quickstart with Hermes
      • Inference Options
      • Use Local Inference
      • Switch Inference Providers
      • Manage Sandbox Lifecycle
      • Runtime Controls
      • Set Up Messaging Channels
      • Workspace Files
      • Backup and Restore
      • Install Hermes Plugins
      • Approve or Deny Network Requests
      • Customize the Network Policy
      • Integration Policy Examples
      • Monitor Sandbox Activity
      • Security Best Practices
      • Credential Storage
      • Architecture Details
      • Commands
      • Which CLI to Use
      • Network Policies
      • Troubleshooting
      • Agent Skills
      • Report Vulnerabilities
      • License
      • Discord
NVIDIANVIDIA
Developer-friendly docs for your API
Privacy Policy | Your Privacy Choices | Terms of Service | Accessibility | Corporate Policies | Product Security | Contact

Copyright © 2026, NVIDIA Corporation.

LogoLogoNemoClaw
On this page
  • Key Features
  • Benefits of Using NemoClaw
  • Use Cases
  • Next Steps
About NemoClaw

Overview of NVIDIA NemoClaw

||View as Markdown|
Previous

NVIDIA NemoClaw

Next

NemoClaw Architecture Overview

NVIDIA NemoClaw is an open-source reference stack for running always-on AI agents more safely inside OpenShell containers. NemoClaw provides onboarding, lifecycle management, and agent operations for supported runtimes in OpenShell sandboxes. It incorporates policy-based privacy and security guardrails, giving you control over your agents’ behavior and data handling. This enables self-evolving agents to run more safely in clouds, on-prem, RTX PCs, and DGX Spark.

NemoClaw pairs hosted models on inference providers or local endpoints with a hardened sandbox, routed inference, and declarative egress policy so deployment stays safer and more repeatable. The sandbox runtime comes from NVIDIA OpenShell. NemoClaw adds the blueprint, nemoclaw CLI, onboarding, and related tooling as the reference way to run supported agents there.

CapabilityDescription
Sandbox supported agentsCreates an OpenShell sandbox pre-configured for your selected agent, with filesystem and network policies applied from the first boot.
Route inferenceConfigures OpenShell inference routing so agent traffic goes to the provider and model you chose during onboarding (NVIDIA Endpoints, OpenAI, Anthropic, Gemini, compatible endpoints, local Ollama, and others). The agent uses inference.local inside the sandbox; credentials stay on the host.
Manage the lifecycleHandles blueprint versioning, digest verification, and sandbox setup.

Key Features

NemoClaw provides the following product capabilities.

FeatureDescription
Guided onboardingValidates credentials, selects providers, and creates a working sandbox in one command.
Agent skillsPackages NemoClaw documentation as user skills so AI coding assistants can guide setup, inference configuration, policy management, monitoring, deployment, security review, and troubleshooting.
Hardened blueprintA security-first Dockerfile with capability drops, least-privilege network rules, and declarative policy.
State managementSafe migration of agent state across machines with credential stripping and integrity verification.
Messaging channelsOpenShell-managed processes connect Telegram, Discord, Slack, and similar platforms to the sandboxed agent. NemoClaw configures channels during onboarding; OpenShell supplies the native constructs, credential flow, and runtime supervision.
Routed inferenceProvider-routed model calls through the OpenShell gateway, transparent to the agent. Supports NVIDIA Endpoints, OpenAI, Anthropic, Google Gemini, compatible endpoints, local Ollama, local vLLM, and the Model Router.
Layered protectionNetwork, filesystem, process, and inference controls that can be hot-reloaded or locked at creation.

Benefits of Using NemoClaw

Autonomous AI agents can make arbitrary network requests, access the host filesystem, and call any inference endpoint. Without guardrails, this creates security, cost, and compliance risks that grow as agents run unattended.

NemoClaw provides the following benefits to mitigate these risks.

BenefitDescription
Sandboxed executionEvery agent runs inside an OpenShell sandbox with Landlock, seccomp, and network namespace isolation. No access is granted by default.
Routed inferenceModel traffic is routed through the OpenShell gateway to your selected provider, transparent to the agent. You can switch providers or models. Refer to Inference Options.
Declarative network policyEgress rules are defined in YAML. Unknown hosts are blocked and surfaced to the operator for approval.
Single CLIThe nemoclaw command orchestrates the full stack: gateway, sandbox, inference provider, and network policy.
Blueprint lifecycleVersioned blueprints handle sandbox creation, digest verification, and reproducible setup.

Use Cases

You can use NemoClaw for various use cases including the following.

Use CaseDescription
Always-on assistantRun a sandboxed agent with controlled network access and operator-approved egress.
Sandboxed testingTest agent behavior in a locked-down environment before granting broader permissions.
Remote GPU deploymentDeploy a sandboxed agent to a remote GPU instance for persistent operation.

Next Steps

Navigate to the following topics to learn more about NemoClaw and how to install and use it.

  • Architecture Overview to understand how NemoClaw works.
  • Ecosystem to understand how your agent, OpenShell, and NemoClaw relate in the wider stack, and when to use NemoClaw versus OpenShell.
  • Quickstart with OpenClaw to install NemoClaw and run your first OpenClaw sandbox.
  • Agent Skills to load NemoClaw guidance into an AI coding assistant.
  • Inference Options to check the inference providers that NemoClaw supports and how inference routing works.