dkr-images-trivy

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install Trivy from Aqua Security's official repository.
  • Evidence: SKILL.md contains commands to fetch the public key from https://aquasecurity.github.io/trivy-repo/deb/public.key and add the repository at https://aquasecurity.github.io/trivy-repo/deb.
  • Context: Aqua Security is a well-known provider of container security tools, and these are the official installation methods for the tool.
  • [PRIVILEGE_ESCALATION]: The documentation includes commands that require administrative privileges for tool installation.
  • Evidence: SKILL.md lists sudo apt-get install and sudo tee commands to manage system packages and repository configurations.
  • Context: This is standard procedure for installing software on Linux systems and is intended for the user to execute manually during setup.
  • [INDIRECT_PROMPT_INJECTION]: The provided Python scripts ingest external image references which are then passed to a CLI tool, representing a surface for potential injection if the inputs are malicious.
  • Ingestion points: The image_ref argument in scripts/agent.py and the image argument in scripts/process.py are taken from user-supplied command-line arguments.
  • Boundary markers: None are explicitly defined in the scripts to delimit the user input within the command execution context.
  • Capability inventory: Both scripts/agent.py and scripts/process.py use subprocess.run to execute the trivy command with the provided image reference.
  • Sanitization: The scripts use command lists (e.g., subprocess.run(["trivy", ...])) which effectively prevents shell-level command injection, though the input is passed directly as an argument to the Trivy tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 11:31 AM