The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the docker CLI to manage container lifecycles and perform debugging, including the ability to execute commands with root privileges inside a container via docker exec -u root.
[DATA_EXFILTRATION]: The skill provides examples of using docker inspect to query container configuration, specifically environment variables (.Config.Env), which frequently contain sensitive information such as API keys, database credentials, and service tokens.
[EXTERNAL_DOWNLOADS]: The skill documents the use of docker pull to download container images from remote registries.
[PROMPT_INJECTION]: The skill processes untrusted external data from container logs and metadata, creating a surface for indirect prompt injection.
Ingestion points: Container logs (docker logs) and container metadata/state (docker inspect) as specified in SKILL.md.
Boundary markers: Absent. No delimiters or instructions are provided to ensure the agent ignores instructions potentially embedded in container logs or labels.
Capability inventory: The skill facilitates high-impact operations including arbitrary command execution (docker exec), resource destruction (docker system prune), and file system access (docker cp).
Sanitization: Absent. The skill does not suggest validation or filtering of the content retrieved from containers before it is processed by the agent.

docker-ops