improve-codebase-architecture

Warn

Audited by Gen Agent Trust Hub on Jul 16, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The generated HTML report in HTML-REPORT.md configures the Mermaid diagramming library with securityLevel: "loose". This setting explicitly disables the library's internal security sandbox, allowing the execution of JavaScript embedded within diagram definitions.
  • [EXTERNAL_DOWNLOADS]: The skill fetches resources from well-known services, including the Tailwind CSS framework and the Mermaid library, via CDN links (tailwindcss.com and jsdelivr.net) embedded in the generated HTML report.
  • [COMMAND_EXECUTION]: The skill uses platform-specific system commands (xdg-open on Linux, open on macOS, and start on Windows) to automatically launch the generated architectural report from the local temporary directory in the user's default web browser.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. It ingests untrusted data from the codebase being scanned (including git commit messages, file contents, and ADRs) and renders it into an HTML report. In the absence of documented sanitization or boundary markers, this data combined with the disabled Mermaid security settings creates a surface for Cross-Site Scripting (XSS) when the report is viewed.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 16, 2026, 04:46 AM
Security Audit — agent-trust-hub — improve-codebase-architecture