The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and use various security tools from external repositories and registries, including Docker Hub, Google Container Registry (GCR), and GitHub.
Evidence: git clone https://github.com/docker/docker-bench-security.git in SKILL.md.
Evidence: curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh in references/build-runtime-security.md.
[REMOTE_CODE_EXECUTION]: The documentation includes recipes for installing and running remote scripts for security analysis, such as the Syft installation script and the Docker Bench Security audit script.
Evidence: curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh | sh in references/build-runtime-security.md.
Evidence: sudo sh docker-bench-security.sh in SKILL.md.
[PRIVILEGE_ESCALATION]: Certain security auditing and monitoring commands require elevated privileges or the use of the --privileged flag to access host-level metrics and configurations.
Evidence: sudo sh docker-bench-security.sh and docker run --privileged ... falcosecurity/falco in SKILL.md.
[DATA_EXFILTRATION]: For the purpose of security scanning, the skill suggests mounting sensitive host paths and the Docker socket into auditing containers.
Evidence: Mounting /var/run/docker.sock, /etc, and /var/lib as read-only volumes for the docker-bench-security tool in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process project files and run external scanning tools (Trivy, Grype, Syft). Maliciously crafted files or container images could potentially output instructions aimed at influencing the agent's behavior during a security review.
Ingestion points: The skill operates on local Dockerfiles and project directories during scanning tasks in SKILL.md.
Boundary markers: Not present.
Capability inventory: Includes docker run, sudo, curl, and git access across SKILL.md and references/build-runtime-security.md.
Sanitization: Not explicitly implemented in the markdown recipes.

docker-security-guide