reviewing-cluster-health
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses official command-line tools for diagnostic purposes, including the
cockroachCLI (e.g.,cockroach node status,cockroach cert list),opensslfor certificate inspection, and cloud provider CLIs (aws ec2,gcloud compute,az vm) for BYOC infrastructure checks. All commands are targeted at cluster health monitoring. - [EXTERNAL_DOWNLOADS]: The skill interacts with the official vendor API at
cockroachlabs.cloudto retrieve cluster state and version information for managed tiers. These requests use user-provided API keys via environment variables and target the vendor's own infrastructure. - [DATA_EXFILTRATION]: Diagnostic data collection is limited to cluster metadata, node status, performance metrics, and configuration settings required for health assessment. No patterns of sensitive file access or unauthorized data transmission to third-party domains were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local command outputs and SQL query results (including
crdb_internalvirtual tables). As these sources are either local system tools or the database itself, the risk of adversarial injection via these channels is low and consistent with standard operational monitoring.
Audit Metadata