deploying-on-gcp
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for building data-driven applications that ingest data from external sources, which is a potential surface for indirect prompt injection if the processed content is later used to influence agent or application logic.
- Ingestion points: Data ingestion points are documented in
references/data-analytics.md(via Pub/Sub messages, Cloud Storage objects, and BigQuery tables) andreferences/ml-ai-services.md(via Vertex AI datasets). - Boundary markers: The provided code templates for data transformation and ML pipelines do not include specific delimiters or instructions to treat ingested data as untrusted or separate from control logic.
- Capability inventory: The skill possesses significant capabilities for environment modification, including provisioning infrastructure via Terraform (
examples/terraform/cloud-run-service.tf), executing shell commands via gcloud (examples/gcloud/common-commands.sh), and deploying executable workloads to Cloud Run, GKE, and Vertex AI. - Sanitization: While the skill emphasizes secure infrastructure practices (e.g., IAM least privilege and VPC Service Controls), the specific Python SDK and Apache Beam templates do not implement content-level sanitization for the data being ingested.
Audit Metadata