k8s-gateway-api-troubleshooting

Pass

Audited by Gen Agent Trust Hub on Nov 8, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses kubectl commands to investigate the status of Kubernetes resources such as GatewayClasses, Gateways, and HTTPRoutes. These are standard administrative actions for troubleshooting. The skill includes explicit instructions to 'Read first and ask before changing anything,' and provides 'Undo' commands for all proposed patches.
  • [DATA_EXPOSURE]: While the skill accesses cluster metadata via kubectl, it includes a mandatory safety rule: 'Never read Secrets: not the TLS certificates that listeners refer to, not CA bundles in Secrets.' This proactively mitigates the risk of sensitive data exposure.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes output from kubectl (JSONPath and YAML). It limits the attack surface by extracting specific status fields and conditions rather than ingesting untrusted blobs of data. Given the primary purpose is Kubernetes administration, the inherent risk is well-managed through the skill's restrictive instructions and user-in-the-loop requirements for any write operations.
  • [SAFE]: The skill provides thorough documentation and source references to official Kubernetes and Gateway API repositories. It distinguishes between fixes that can be requested from the user ('Ask first') and those that the agent should never run autonomously ('Proposed fix'), ensuring that shared cluster infrastructure is protected from unauthorized automated modification.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 8, 2026, 10:57 PM