improve-codebase-architecture
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches Tailwind CSS and the Mermaid.js library from well-known content delivery networks (tailwindcss.com and jsdelivr.net) to provide styling and diagramming for the architectural report.
- [COMMAND_EXECUTION]: It uses platform-specific commands including xdg-open, open, and start to automatically launch the generated architectural review file for the user.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the report generation process. It uses Mermaid.js with the securityLevel set to 'loose', which permits the execution of HTML and scripts. If an analyzed codebase contains malicious content in its metadata or file structure, that content could be rendered and executed in the user's browser when the report is opened.
- Ingestion points: Codebase contents, project context (CONTEXT.md), and architectural decision records (ADRs) processed by the Explore subagent.
- Boundary markers: No explicit delimiters or instructions are provided to the agent to sanitize or ignore instructions embedded within the codebase data before placing it in the HTML template.
- Capability inventory: The skill has the ability to write to the OS temporary directory and execute shell commands to open those files.
- Sanitization: The instructions do not define a process for escaping or validating codebase content prior to its inclusion in the HTML report.
Audit Metadata