metro-ai-apps-recipe

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill scaffolds and executes several utility scripts (install.sh, validate_env.sh, sample_start.sh, sample_watchdog.sh) to manage the Docker lifecycle. These scripts are used for environment preparation, pipeline launching, and health monitoring.
  • [EXTERNAL_DOWNLOADS]: Fetches sample videos and model weights from the author's official GitHub repositories and downloads standard software packages via npm and pip. These downloads are from trusted and well-known sources.
  • [PRIVILEGE_ESCALATION]: The skill provides instructions for mounting the host's Docker socket (/var/run/docker.sock) into a container as a documented fallback for running tests on hosts lacking local development tools. It also configures specific device access (/dev/dri) for hardware acceleration.
  • [INDIRECT_PROMPT_INJECTION]: The skill handles untrusted external data from RTSP video streams and MQTT detection metadata. While it includes data coercion and filtering logic in Node-RED function nodes, this ingestion point creates a potential surface for indirect injection via manipulated video data or metadata strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 12:05 AM
Security Audit — agent-trust-hub — metro-ai-apps-recipe