improve-codebase-architecture
Pass
Audited by Gen Agent Trust Hub on Jul 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill generates an HTML report that loads resources from well-known external CDNs: Tailwind CSS from cdn.tailwindcss.com and Mermaid.js from cdn.jsdelivr.net.
- [COMMAND_EXECUTION]: The skill uses shell commands (xdg-open, open, or start) to open a generated HTML report in the user's default browser.
- [PROMPT_INJECTION]: The skill ingests untrusted content from the local codebase (source files, ADRs, and glossary). Ingestion points: CONTEXT.md, docs/adr/, and files explored via sub-agent. Capability inventory: file write to temporary directory and shell command execution. Sanitization: No explicit sanitization or boundary markers defined for interpolated content.
- [REMOTE_CODE_EXECUTION]: The generated HTML report configures the Mermaid diagram renderer with securityLevel: 'loose'. This setting allows scripts embedded in diagram definitions to execute. If an attacker manages to inject a malicious diagram string into a source file or ADR that the AI then extracts into the report, it could lead to arbitrary JavaScript execution in the context of the locally opened HTML file.
Audit Metadata