aws-architecture-diagram
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill analyzes local infrastructure code (CDK, Terraform, CloudFormation), which presents a surface for indirect prompt injection. This is a low-risk finding inherent to codebase scanning features. Evidence chain: 1. Ingestion points: Codebase scanning for infrastructure definitions in Mode A (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Bash, Write, Read, Glob, and Grep tools (SKILL.md). 4. Sanitization: Absent.
- [COMMAND_EXECUTION]: Executes local Python scripts within the plugin's directory structure to fix geometry overlaps and generate browser-based preview URLs.
- [EXTERNAL_DOWNLOADS]: Recommends the installation of the legitimate and well-known
defusedxmlpackage from the official Python Package Index (PyPI) to prevent XML entity expansion attacks.
Audit Metadata