The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data to generate and suggest CLI commands, creating an injection surface.
Ingestion points: User queries referencing build instructions, platform names, or Dockerfile paths as defined in the 'When to Use' section of SKILL.md.
Boundary markers: The skill does not provide instructions to the agent to treat user-supplied strings as data rather than instructions, nor does it use delimiters to isolate user input.
Capability inventory: The skill facilitates execution of the docker and docker buildx toolsets, which include file system access and network operations (e.g., docker push, docker buildx build --push).
Sanitization: There is no evidence of validation or sanitization of user-provided arguments before they are interpolated into suggested commands.
[COMMAND_EXECUTION]: The skill provides instructions for executing complex shell commands involving Docker build orchestration and SSH agent management.
Evidence: SKILL.md and various files in examples/ demonstrate commands for building images, mounting secrets, and forwarding SSH keys (e.g., eval $(ssh-agent) && ssh-add).
[PRIVILEGE_ESCALATION]: The skill references commands that require elevated host permissions to configure the container runtime environment.
Evidence: In references/03-multi-platform-strategies.md, the instruction docker run --privileged --rm tonistiigi/binfmt --install all uses the --privileged flag, which grants the container broad access to host kernel features (binfmt_misc) and devices.

docker-buildx