The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local codebase environment, including commit history via git log --oneline and contents of source files, to generate an architectural report. There are no explicit instructions to sanitize this content before it is interpolated into the HTML report, creating a path for malicious strings in the codebase to influence the report output.
[DYNAMIC_EXECUTION]: In HTML-REPORT.md, the Mermaid.js library is initialized with securityLevel: "loose". This setting allows the library to render raw HTML and potentially execute scripts embedded within diagram definitions. If malicious content from the codebase is included in a diagram description, it could lead to code execution in the context of the user's browser when the report is opened.
[COMMAND_EXECUTION]: The skill is instructed to automatically open the generated report using system-specific commands such as xdg-open, open, or start. While this is a common feature for report generation, the lack of sanitization of the file's content increases the risk associated with automatic execution.
[EXTERNAL_DOWNLOADS]: The generated report pulls styling and graphing libraries from cdn.tailwindcss.com and cdn.jsdelivr.net. These are well-known and reputable services, used here to facilitate the visual presentation of the architectural analysis.

improve-codebase-architecture