improve-codebase-architecture

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches Tailwind CSS and Mermaid.js from external CDNs (cdn.tailwindcss.com and cdn.jsdelivr.net). These are well-known services used for layout and diagramming purposes.
  • [COMMAND_EXECUTION]: The skill uses system commands (xdg-open, open, or start) to automatically open a generated HTML report stored in the system's temporary directory.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting codebase content and rendering it in a browser-based HTML report. The Mermaid.js diagramming tool is explicitly configured with securityLevel: "loose", which allows rendering of HTML tags within diagrams. This configuration, combined with the lack of explicit sanitization of the analyzed code snippets, creates a potential attack surface for indirect injection (XSS) if a malicious actor places payloads in code comments or documentation that the skill then processes.
  • Ingestion points: Codebase files (functions, classes, packages) explored by the agent via the subagent tool.
  • Boundary markers: None. The instructions do not specify using delimiters or warnings for the content rendered in the HTML report.
  • Capability inventory: File system read (codebase), file system write (temporary HTML report), and shell execution (opening the report).
  • Sanitization: Absent. The instructions do not provide guidelines for escaping or validating codebase content before it is interpolated into the HTML template.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 11:31 AM
Security Audit — agent-trust-hub — improve-codebase-architecture