improve-codebase-architecture
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill generates an HTML report that loads Tailwind CSS and Mermaid.js from well-known, trusted CDNs (tailwindcss.com and jsdelivr.net). These are standard frontend development practices for visualization and do not pose a security risk in this context.
- [COMMAND_EXECUTION]: The skill uses
git log --onelineto identify codebase 'hot spots' and utilizes platform-specific commands (xdg-open,open,start) to display the generated architectural report to the user. These actions are aligned with the skill's primary purpose of codebase analysis. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase data (source code, commit history, ADRs) and interpolates it into a visual report.
- Ingestion points: Reads local files and git history via sub-agents.
- Boundary markers: None explicitly defined in the instructions for data interpolation.
- Capability inventory: File system read access, temporary file writing, and local command execution for opening the report.
- Sanitization: The skill relies on standard LLM processing and uses a static HTML scaffold, which limits the risk of executable injection within the report context.
Audit Metadata