gcp-cloud-storage
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands using
gcloud storage,gsutil, andjqto interact with Google Cloud Storage. These commands are used for legitimate administrative tasks such as listing buckets, describing configurations, auditing IAM policies, and checking storage statistics. - [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface where it processes metadata from external cloud resources (bucket names, labels, IAM policy members, and storage metrics).
- Ingestion points: Commands like
gcloud storage buckets listandgcloud storage lspull data from the GCP environment into the agent's context. - Boundary markers: None explicitly defined to separate resource data from instructions.
- Capability inventory: The skill uses CLI tools for read operations and monitoring.
- Sanitization: Results are parsed through
jqto extract specific fields, reducing the risk of processing raw malicious strings as instructions, though the agent remains responsible for interpreting the structured output safely.
Audit Metadata