k8s-fluent-bit-troubleshooting

Pass

Audited by Gen Agent Trust Hub on Nov 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by reading container logs (kubectl logs) in SKILL.md and reference files. An attacker with control over application logs could attempt to embed instructions to influence the agent's behavior.
  • Ingestion points: kubectl logs commands used for troubleshooting in SKILL.md, references/config-and-operator.md, and references/pipeline.md.
  • Boundary markers: None identified; logs are processed as raw text without specific delimiters or isolation instructions.
  • Capability inventory: The skill can perform kubectl apply and kubectl rollout restart (proposed fixes), though it requires user approval.
  • Sanitization: No sanitization or filtering of log content is specified to prevent command or prompt injection.
  • [COMMAND_EXECUTION]: The skill utilizes kubectl to query and manage Kubernetes resources. The skill includes robust safety instructions, such as explicitly forbidding the agent from reading Secrets or repeating credentials found in ConfigMaps, and requiring user approval for any modification to the cluster state.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 10, 2026, 11:19 PM