The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill includes hardcoded sensitive information within Kubernetes Secret and Docker Compose examples, which may lead to insecure deployments if used without modification.
Evidence: Hardcoded DATABASE_URL: "postgresql://user:pass@db:5432/app" and API_KEY: "secret-api-key" in SKILL.md.
[COMMAND_EXECUTION]: The skill requests powerful tools and provides instructions for interactive shell access to running containers, which increases the risk of unauthorized system manipulation.
Evidence: Permission for the Bash tool is requested in the frontmatter.
Evidence: Instructions for docker exec -it and kubectl exec -it allow arbitrary command execution inside container environments.
Evidence: The 'Multi-Stage Build Benefits' section contains multiple null bytes (\0) and control characters (\x0C, \x10, \x02, \x14, \x18), which is a form of obfuscation that may be used to evade static analysis or security filters.

docker-kubernetes