The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the gcloud and kubectl command-line tools to manage and analyze Google Kubernetes Engine clusters. These are standard operations for GKE administration, including cluster discovery, node pool management, and workload status checks.
[INDIRECT_PROMPT_INJECTION]: The skill ingests resource names and metadata from the GKE environment via CLI output. It mitigates potential injection risks by using jq to extract and structure specific data fields into JSON format before presentation, ensuring the agent processes filtered data rather than raw, potentially malicious strings.
[SAFE]: The skill incorporates robust administrative practices, such as discovery-first rules and anti-hallucination guidelines. These instructions ensure that the agent verifies the existence of clusters and node pools before performing actions, preventing errors or unintended operations based on assumed state.

gcp-gke