The Agent Skills Directory

[COMMAND_EXECUTION] (LOW): The skill contains commands intended to be executed by the agent to interact with Google Cloud Platform. These are standard GKE management operations (gcloud container clusters create, gcloud iam service-accounts).
[DATA_EXPOSURE] (SAFE): No hardcoded credentials, API keys, or unauthorized file access patterns were found. It uses placeholders like ${PROJECT_ID} for environment-specific data.
[REMOTE_CODE_EXECUTION] (SAFE): No scripts are downloaded from external sources, and there is no usage of curl | bash or similar dangerous patterns.
[PROMPT_INJECTION] (SAFE): No instructions designed to override agent behavior or bypass safety guardrails were identified.
[INDIRECT_PROMPT_INJECTION] (SAFE): The skill does not ingest untrusted external data from web sources or files, minimizing the risk of indirect injection attacks.

gcp-gke