juicebox-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 shell commands such as
curl,jq,nslookup, andkubectlto perform system diagnostics, check API status, and modify deployment environments during incident response. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external data from the Juicebox API (health status, search results, account quotas) and Kubernetes logs. It includes explicit instructions to redact sensitive information, such as contact details, enrichment values, and credentials, before producing the final incident receipt.
- [SAFE]: The skill uses environment variables for secret management (e.g.,
JUICEBOX_API_KEY) and communicates with official vendor domains (api.juicebox.ai), which aligns with standard security best practices for DevOps automation.
Audit Metadata