The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from the Kubernetes cluster (such as pod logs, resource status conditions, and cluster events) which could potentially contain malicious instructions aimed at the agent. This attack surface is a necessary part of the skill's primary function as a debugger.
Ingestion points: The agent retrieves diagnostic data using tools like get_kubernetes_resources, get_kubernetes_logs, and get_kubernetes_events (documented in SKILL.md).
Boundary markers: The instructions do not explicitly require the use of unique delimiters or "ignore" instructions when processing data retrieved from the cluster tools.
Capability inventory: The skill possesses extensive cluster modification capabilities including apply_kubernetes_manifest, patch_kubernetes_resource, delete_kubernetes_resource, and reconcile_flux_resource (documented in SKILL.md).
Sanitization: No specific sanitization or filtering of the retrieved cluster data is requested in the instructions, though the skill notes that Kubernetes Secret values are masked by the underlying tools.
[COMMAND_EXECUTION]: The skill instructs the agent to execute local shell commands for building and diffing manifests during local development workflows.
Evidence: Workflow 9 in SKILL.md specifies the use of kustomize build <path> --load-restrictor=LoadRestrictionsNone, flux-operator build resourceset, and helm template to preview changes locally.

gitops-cluster-debug