The Agent Skills Directory

Standard Command Execution: The skill utilizes gcloud and kubectl to retrieve diagnostic information from the GKE cluster, such as pod status, events, and application logs. These commands are limited to the skill's intended purpose of infrastructure diagnostics.
Usage Attribution and Telemetry: The skill includes instructions to include attribution tags in gcloud commands and HTTP headers (User-Agent). According to the documentation, this is used for metrics and does not transmit sensitive user data.
Indirect Prompt Injection Surface: The skill ingests untrusted data from the cluster environment, such as application logs (kubectl logs) and Kubernetes events (kubectl get events). This represents an attack surface where malicious content embedded in logs could potentially attempt to influence the agent's reasoning.
Ingestion points: kubectl logs, kubectl get events, and gcloud logging read command outputs.
Boundary markers: No explicit delimiters are used to wrap external log content.
Capability inventory: The skill can execute shell commands (gcloud, kubectl) and perform Git operations (git branch, git commit, git push) to create Pull Requests.
Sanitization: No specific sanitization or filtering of the log content is described before analysis.
GitOps Integration: The skill is designed to interact with GitHub to propose manifest corrections via Pull Requests. This workflow ensures that changes are reviewable by a human operator before being merged into the codebase.

gke-workload-troubleshooting