The Agent Skills Directory

[SAFE]: The skill implements comprehensive defensive prompt engineering designed to mitigate common container-related security risks. It explicitly forbids exfiltrating container logs or environment variables to external network endpoints.
[SAFE]: Instructions are provided to redact sensitive information such as API keys and passwords from Docker command outputs before they are displayed to the user.
[SAFE]: The skill requires explicit user confirmation for potentially destructive operations (rm, prune) and for operations involving external registries (push, pull from unverified sources).
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes output from container logs and metadata which could contain untrusted instructions.
Ingestion points: docker logs and docker inspect output processed by the agent (SKILL.md).
Boundary markers: Employs a 'Security Guardrails' section with explicit constraints on behavior when handling sensitive data.
Capability inventory: Uses the exec tool to run arbitrary docker CLI commands (SKILL.md).
Sanitization: Includes specific instructions to redact secret values (passwords, tokens) and summarize large outputs.

Docker Manager-hardened