google-cloud-solution-agentic-ai-bidirectional-streaming

Pass

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • External Resource Retrieval: The skill is designed to retrieve architectural guidance and implementation details from official Google Cloud documentation domains (docs.cloud.google.com), developer resources (adk.dev, codelabs.developers.google.com), and official GitHub repositories (github.com/gca-americas). This process is used to ground the agent's output in official technical recommendations.\n- Code and Script Generation: The skill generates Infrastructure as Code (Terraform) and verification scripts (using gcloud and curl) for the user to deploy and validate their solution. To maintain security, the workflow incorporates multiple human-in-the-loop checkpoints (Phase 2 Step 6, Phase 3 Step 5, and Phase 4 Step 5) where the user is prompted to review and approve all generated content before proceeding.\n- Indirect Prompt Injection Surface: Since the skill processes user-defined requirements and external documentation to generate technical guidance, there is a theoretical surface for indirect prompt injection. This is addressed by the skill's iterative design, which requires explicit user confirmation of the technical decomposition and architecture before code is generated.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 7, 2026, 05:21 PM
Security Audit — agent-trust-hub — google-cloud-solution-agentic-ai-bidirectional-streaming