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

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it collects functional and non-functional requirements from the user and subsequently generates executable code, such as Terraform configurations and shell scripts, based on that input.
  • Ingestion points: User-provided requirements gathered during Phase 1: Requirements discovery and analysis in SKILL.md.
  • Boundary markers: There are no specific instructions or delimiters provided to the agent to distinguish between user input and system instructions during the code generation process.
  • Capability inventory: The agent is instructed to write to solution-architecture-guide.md and generate Infrastructure as Code (IaC) and verification scripts in Phase 3 and Phase 4.
  • Sanitization: The skill does not include instructions for validating or sanitizing user-provided strings before they are incorporated into generated scripts or configurations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 05:40 AM
Security Audit — agent-trust-hub — google-cloud-solution-agentic-ai-bidirectional-streaming