isaac-mission-control-showcase

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill extensively uses subprocess.run() and Bash scripts to automate the robotics stack. It executes git for dependency management, docker for container orchestration, and various host diagnostic tools like nvidia-smi and ss. All command invocations use list-based arguments or validated inputs to prevent shell injection.
  • [EXTERNAL_DOWNLOADS]: Runtime dependencies are resolved from official NVIDIA GitHub organizations (nvidia-isaac, isaac-sim) and assets are fetched from an NVIDIA-owned S3 bucket (omniverse-content-production.s3-us-west-2.amazonaws.com). These sources are trusted vendor resources required for the skill's functionality.
  • [DYNAMIC_EXECUTION]: The skill dynamically resolves entrypoints for upstream skills using a locked version manifest (upstream-versions.lock.json). It also executes Python scripts within a local Isaac Sim instance via a remote server adapter (isaac-sim-remote), which is the standard mechanism for controlling the simulator.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests state data from Mission Database APIs and Isaac Sim server responses. This is mitigated by robust sanitization, including the use of shlex.quote for shell arguments and regex validation for user-supplied robot identities.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 02:03 PM
Security Audit — agent-trust-hub — isaac-mission-control-showcase