groq-incident-runbook

Warn

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The runbook contains a command (kubectl get secret groq-secrets -o jsonpath='{.data.api-key}' | base64 -d) that retrieves, decodes, and prints a sensitive API key to the agent's output, exposing credentials in plain text.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of external, untrusted data.
  • Ingestion points: Kubernetes log streams via kubectl logs and HTTP responses from curl https://status.groq.com.
  • Boundary markers: Absent. No delimiters or security instructions are provided to the agent to treat this data as untrusted.
  • Capability inventory: The agent has access to kubectl and curl, which can be misused if the agent is manipulated by instructions hidden in the ingested data.
  • Sanitization: None. Ingested data is passed directly to the agent's context.
  • [COMMAND_EXECUTION]: The skill executes multiple administrative commands, including modifying deployment configurations (kubectl set env), performing rollout restarts, and running local diagnostic scripts (./scripts/groq-debug-bundle.sh).
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 12, 2026, 12:34 AM