k8s-upgrade-readiness

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes live data from a Kubernetes cluster which serves as an untrusted input surface.
  • Ingestion points: The skill executes numerous kubectl get and kubectl --raw /metrics commands across SKILL.md, references/drains.md, and references/versions-and-apis.md to inspect the cluster state.
  • Boundary markers: The instructions do not define clear boundaries, delimiters, or warnings for the agent to ignore potential instructions that could be embedded in resource names, labels, or metric values retrieved from the cluster.
  • Capability inventory: The agent is instructed to execute shell commands and kubectl operations which could be influenced by malicious cluster data.
  • Sanitization: There is no requirement or logic provided to sanitize or validate the content of the data retrieved from the Kubernetes API before the agent interprets or reports it.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 03:37 AM