The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references Tailwind CSS and Mermaid.js from well-known services (cdn.tailwindcss.com and cdn.jsdelivr.net) to style and generate diagrams for its generated HTML audit reports.
[COMMAND_EXECUTION]: To show results to the user, the skill uses platform-specific commands (open, xdg-open, or start) to automatically launch the generated HTML report file in the user's browser.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection because it ingests untrusted codebase data while maintaining tool-calling and file-writing capabilities. * Ingestion points: The skill reads git log history, Architecture Decision Records (ADRs), and various source code files during exploration. * Boundary markers: There are no instructions defining specific delimiters to separate external codebase content from the agent's internal logic and instructions. * Capability inventory: The skill can write PRD files to the repository, call architectural design and ticket-slicing tools, and execute shell commands to open files. * Sanitization: The skill does not mention sanitizing, escaping, or validating the content read from the codebase before processing it.

architecture-audit