improve-codebase-architecture
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted codebase data and renders it into a self-contained HTML report. The Mermaid.js configuration uses securityLevel: 'loose', which permits HTML/JS tags and could allow malicious content embedded in the codebase to execute when the report is viewed. * Ingestion points: Codebase files explored via the Explore subagent. * Boundary markers: None provided to prevent the agent from obeying instructions in the code. * Capability inventory: Ability to write to temp directory and execute OS commands to open files. * Sanitization: No explicit sanitization of codebase content before insertion into the report template.
- [EXTERNAL_DOWNLOADS]: The generated report fetches the Tailwind CSS framework and Mermaid.js library from well-known public CDNs (tailwindcss.com and jsdelivr.net).
- [COMMAND_EXECUTION]: The skill executes operating system commands (xdg-open, open, or start) to automatically open the generated architectural report for the user.
Audit Metadata