nemoclaw-get-started

Pass

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Fetches and executes installation and uninstallation shell scripts from 'www.nvidia.com' and the official 'NVIDIA' GitHub organization. These resources are verified vendor assets intended for environment setup.
  • [COMMAND_EXECUTION]: Performs local configuration tasks, including the modification of shell profiles (~/.bashrc, ~/.zshrc) to update the PATH environment variable and the execution of the nemoclaw and openclaw command-line utilities.
  • [PROMPT_INJECTION]: The skill enables interaction with an AI agent, creating a surface for potential indirect prompt injection.
  • Ingestion points: The openclaw agent command accepts arbitrary text input via the -m flag.
  • Boundary markers: The agent is isolated within a sandbox utilizing Landlock, seccomp, and network namespaces.
  • Capability inventory: The sandboxed environment allows for tool execution and network operations as defined by security policies.
  • Sanitization: No explicit input sanitization or filtering is described in the installation documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 7, 2026, 04:30 AM
Security Audit — agent-trust-hub — nemoclaw-get-started