The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes capabilities to ingest and process potentially untrusted data from cluster environments, which presents a standard surface for indirect prompt injection attacks.
Ingestion points: Tools such as get_k8s_logs, get_k8s_resource, and list_k8s_events (referenced in references/cli-reference.md and references/mcp-usage.md) fetch logs, metadata, and event descriptions from running workloads.
Boundary markers: The instructions do not define specific delimiters or "ignore" instructions for the agent when interpreting the output of these diagnostic tools.
Capability inventory: The skill allows for significant cluster modifications via apply_k8s_manifest, resource patching, and provides shell access through fallback commands like kubectl exec and kubectl run (documented in references/cli-reference.md).
Sanitization: There is no explicit logic described to sanitize or validate the content of pod logs or resource descriptions before they are processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill references official Kubernetes and Google Cloud client libraries.
Fetches the kubernetes Python package and @kubernetes/client-node Node.js package from official registries (referenced in references/client-library-usage.md).
References the official client-go repository from the k8s.io domain.
[SAFE]: Documentation and infrastructure-as-code examples utilize trusted domains and official repositories, including cloud.google.com, hashicorp.com, and terraform.io (referenced in references/iac-usage.md and references/mcp-usage.md).

gke-basics