intercom-incident-runbook
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
kubectl rollout restartto remediate service failures by restarting application deployments. While this modifies the active state of the environment, it is scoped to the incident response workflow and is explicitly authorized in the skill configuration. - [COMMAND_EXECUTION]: The mitigation instructions include
aws secretsmanager update-secretfor rotating compromised Intercom tokens. This is a secure alternative to hardcoding credentials and follows industry best practices for secret management. - [INDIRECT_PROMPT_INJECTION]: The triage process ingests data from external APIs to drive diagnostic decisions. Ingestion points: API responses from
status.intercom.comand local application health endpoints are processed inreferences/triage.md. Boundary markers: None present. Capability inventory: The skill possesses capabilities for network requests (curl), deployment management (kubectl), and secret modification (aws cli). Sanitization: The skill usesjqto parse the structure of JSON responses but does not include validation logic for the content within those fields.
Audit Metadata