The Agent Skills Directory

[External Downloads]: The skill references installation methods for third-party security tools. It includes a command pattern to download and execute an installation script for Trivy from the official Aqua Security GitHub repository. This is a common practice for this tool, but users should verify the source before execution.
[Command Execution]: The agent is instructed to execute local commands for auditing dependencies and running a pre-flight configuration check using a local Python module. These commands are necessary for the skill's primary function of system health checking.
[Indirect Prompt Injection Considerations]: The skill processes external, untrusted content from vulnerability databases, including package descriptions and advisory notes. The ingestion points are the JSON outputs of the audit tools. The skill uses explicit 'Golden Rules' as boundary markers to ensure this content is treated as data evidence rather than instructions. Its capabilities include running subprocesses for auditing and repository cloning, but it emphasizes read-only operations and instructs the agent to avoid following commands embedded in external data.

dependency-audit