skills/googlecloudplatform/gemini-cloud-assist-mcp/designing-and-deploying-infrastructure/Gen Agent Trust Hub
designing-and-deploying-infrastructure
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill performs an exhaustive analysis of user-provided application codebases, including source code, dependency files (requirements.txt, package.json), and configuration files (nginx configs, CI/CD YAMLs). This information is used to populate prompts for the
gemini_cloud_assist:design_infratool. An attacker could embed malicious instructions in code comments or configuration values to influence the infrastructure design or deployment outcome. - Ingestion points: Deep scanning of application source code, dependency files, and configuration files.
- Boundary markers: The skill lacks explicit instructions to use delimiters or ignore embedded instructions when processing the codebase.
- Capability inventory: The skill can design infrastructure, manage application templates, and deploy resources using GCP-specific tools.
- Sanitization: No explicit sanitization or filtering of the codebase content is mentioned before it is passed to the design tools.
- [COMMAND_EXECUTION]: The skill instructions mandate that the agent build and upload container images to the Google Artifact Registry if they are not already present. This requires executing shell commands for containerization and registry operations on the host system.
Audit Metadata