speak-incident-runbook
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as service logs and API responses. An attacker could potentially attempt to embed malicious instructions in these data sources to influence the agent's behavior during triage.\n
- Ingestion points: Triage scripts in references/implementation-guide.md ingest data from kubectl logs and external status APIs.\n
- Boundary markers: None explicitly defined in the triage scripts.\n
- Capability inventory: The skill documentation references the use of kubectl, curl, npm, and gsutil to manage cloud infrastructure and application state.\n
- Sanitization: No explicit sanitization of log content or API responses is shown before analysis.\n- [DATA_EXFILTRATION]: The implementation guide includes administrative commands to inspect Kubernetes secrets (kubectl get secret speak-secrets) to verify that the Speak API credentials are correctly configured.\n- [COMMAND_EXECUTION]: The skill provides scripts that execute standard infrastructure management commands (kubectl, curl, gsutil) to perform health checks, analyze logs, and query performance metrics.
Audit Metadata