gcp-serverless-tf
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project files to infer infrastructure requirements, creating an attack surface for indirect prompt injection.
- Ingestion points: Project configuration files (Dockerfile, main.py, etc.) are read in Step 2 of
SKILL.mdto identify required GCP services. - Boundary markers: None explicitly defined in the skill instructions to separate file content from the agent's logic.
- Capability inventory: File generation/writing and information retrieval via documented MCP tools.
- Sanitization: None mentioned; the skill relies on the agent's internal safety filters and manual user review (e.g.,
terraform plan) of the generated output. - [SAFE]: The skill provides guidance for setting up official MCP tools from Google and Context7, ensuring access to the latest documentation through secure integration methods.
- [SAFE]: Terraform templates in
references/resource-map.mdfollow security principles like least-privilege IAM roles and variable-based configuration to avoid hardcoded secrets.
Audit Metadata