gcp-cloud-run
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from Google Cloud Run logs and service metadata, which constitutes a potential indirect prompt injection attack surface.\n
- Ingestion points: The
gcloud logging readandgcloud run services describecommands inSKILL.mdingest data from external logs and cloud metadata.\n - Boundary markers: No specific delimiters or boundary instructions are used to isolate external data from the agent's control context.\n
- Capability inventory: The skill performs read-only
gcloudoperations to retrieve service status, logs, IAM policies, and monitoring metrics.\n - Sanitization: The skill employs
jqto parse and extract specific fields from JSON command outputs, which reduces the risk of processing unfiltered content.\n- [COMMAND_EXECUTION]: The skill executes gcloud CLI commands to interact with Google Cloud services.\n - Evidence: Commands such as
gcloud run services list,gcloud logging read, andgcloud monitoring time-series listare used to interact with cloud infrastructure and retrieve data.
Audit Metadata