k8s-cert-manager-troubleshooting
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
kubectlcommands to investigate and resolve issues within a Kubernetes cluster. This includes read-only operations likeget,describe, andlogs, as well as modification commands likepatchandscale. The skill explicitly categorizes modification commands as requiring user approval. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes potentially untrusted data from the Kubernetes API, which could be influenced by external entities such as ACME servers or cluster logs.
- Ingestion points:
kubectl logs,kubectl describe, andkubectl eventsare used across SKILL.md and reference files to gather troubleshooting data. - Boundary markers: The skill relies on standard command output presentation; no specific delimiter-based isolation for the data is defined in the instructions.
- Capability inventory: The skill possesses the ability to modify cluster resources via
kubectl patch(to change issuer types) andkubectl scale(to adjust deployment replicas). - Sanitization: No explicit sanitization or filtering of the command output is performed before the agent interprets the results.
Audit Metadata