The Agent Skills Directory

[COMMAND_EXECUTION]: The generated Makefile includes a shell-root target that executes docker compose exec -u root, providing root user access within the container environment.
[COMMAND_EXECUTION]: The template includes a docker-prune target that runs docker system prune -f, a high-impact command that removes unused Docker resources system-wide, potentially affecting other projects on the same host.
[PROMPT_INJECTION]: The skill instructions create an attack surface for indirect prompt injection by processing untrusted local project files.
Ingestion points: The agent is directed to read and analyze local docker-compose.yml files and project directory structures.
Boundary markers: No boundary markers or specific instructions are provided to the agent to distinguish between project metadata and embedded malicious instructions.
Capability inventory: The skill generates a Makefile with targets that execute shell commands for container lifecycle, database management, and system maintenance.
Sanitization: There is no specified logic to sanitize framework names or service names detected in the environment before they are included in the generated file.

create-docker-makefile