improve-codebase-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill uses local system commands (
git log,xdg-open,open,start) to analyze repository history and open generated reports. These are standard development tools and are used as intended for a local agent. - [EXTERNAL_DOWNLOADS]: The generated HTML report fetches the Tailwind CSS and Mermaid.js libraries from well-known content delivery networks (cdn.tailwindcss.com and cdn.jsdelivr.net). These are standard, reputable sources for web assets.
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect injection because it ingests untrusted codebase data and renders it in a browser-based report.
- Ingestion points: The agent traverses the local codebase, reading files to identify architectural patterns (SKILL.md, Step 1).
- Boundary markers: No specific sanitization or escaping markers are defined for codebase content interpolated into the HTML report.
- Capability inventory: The skill writes a temporary HTML file and triggers the system's default browser to open it (SKILL.md, Step 2).
- Sanitization: The HTML report template explicitly configures Mermaid.js with
securityLevel: "loose"(HTML-REPORT.md). This setting allows diagrams to execute scripts. If the codebase being analyzed contains malicious Mermaid syntax (e.g., in code comments), the agent might include it in the report, potentially leading to Cross-Site Scripting (XSS) when the user views the findings.
Audit Metadata