managing-cluster-capacity

Pass

Audited by Gen Agent Trust Hub on Sep 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill facilitates cluster management through various command-line tools.
  • Evidence: Includes commands for the cockroach binary to drain, decommission, and monitor nodes, as well as cloud-specific tools (aws ec2, gcloud compute, az vm) for BYOC infrastructure verification.
  • [EXTERNAL_DOWNLOADS]: The skill interacts with the vendor's official cloud services to manage managed cluster tiers.
  • Evidence: Uses curl to send PATCH requests to the Cockroach Labs Cloud API at https://cockroachlabs.cloud/api/v1/clusters/ for scaling operations.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes system status and job metadata which represents a theoretical surface for indirect injection.
  • Ingestion points: Reads output from commands like cockroach node status and SQL queries against SHOW JOBS or internal crdb_internal metadata tables.
  • Boundary markers: The instructions do not explicitly define delimiters for command output, relying on standard agent handling of tool responses.
  • Capability inventory: The skill has capabilities to execute shell commands, run SQL, and interact with the vendor's web API.
  • Sanitization: No specific filtering or sanitization of database metadata is required or provided within the skill instructions, as the data source is the system being managed.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 29, 2026, 03:34 PM