skills/cloudvoyant/codevoyant/docker/Gen Agent Trust Hub

docker

Pass

Audited by Gen Agent Trust Hub on Jun 28, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches the mise installation script from https://mise.run and base images from official registries including Docker Hub, GCP Artifact Registry, and Microsoft Artifact Registry.
  • [REMOTE_CODE_EXECUTION]: Executes the mise installation script via shell piping (curl | sh) during container and environment setup. This aligns with the documented installation procedure for the mise tool.
  • [COMMAND_EXECUTION]: Contains recipes that perform shell-based operations for container management, GCP authentication, and Cloud Run deployment using docker, gcloud, and mise CLI tools.
  • [PROMPT_INJECTION]: Features an indirect prompt injection surface through the ingestion of external project files.
  • Ingestion points: Accesses version.txt and apps/web/package.json to configure build versions and runtime dependencies.
  • Boundary markers: External data is interpolated directly into commands without isolation markers.
  • Capability inventory: The execution environment permits subprocess calls for container and cloud management tools.
  • Sanitization: File content is used directly in tasks without prior validation or escaping.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 28, 2026, 04:54 PM
Security Audit — agent-trust-hub — docker