google-cloud-solution-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user requirements to drive architectural recommendations and validation logic.
- Ingestion points: Phase 1 in
SKILL.mdcollects functional and non-functional requirements directly from the user. - Boundary markers: Absent. The skill does not instruct the agent to use delimiters or ignore instructions embedded within the user's requirement descriptions.
- Capability inventory: The skill performs shell execution (
terraform plan,gcloud,curl,ping) in Tasks 3.1 and 3.2 ofSKILL.md, and writes code files to the workspace in Phase 4. - Sanitization: Absent. There is no logic provided to sanitize user input before it influences the generated commands or files.
- [COMMAND_EXECUTION]: The skill initiates shell commands for infrastructure validation.
- Evidence:
SKILL.md(Task 3.1 and 3.2) explicitly directs the agent to executeterraform plan,gcloud --dry-run,curl, andping. While intended for architectural validation, this capability is triggered by logic derived from user-supplied data.
Audit Metadata