The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git log to analyze commit history for codebase hotspots. It also uses system-specific commands (xdg-open, open, or start) to launch the generated HTML report for the user.
[EXTERNAL_DOWNLOADS]: The architectural reports generated by the skill reference external libraries (Tailwind CSS and Mermaid.js) from well-known content delivery networks (Tailwind CDN and jsDelivr). These are standard frontend dependencies for rendering the visual report.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes external, potentially untrusted data from the codebase to inform its architectural review, creating a surface for indirect injection.
Ingestion points: Reads project source code, git commit messages, CONTEXT.md files, and Architectural Decision Records (ADRs) within the analyzed repository.
Boundary markers: None identified in the prompt instructions to distinguish between internal instructions and ingested codebase content.
Capability inventory: Capable of writing files to the system temporary directory (HTML reports), modifying or creating markdown files in the repository (CONTEXT.md, ADRs), and executing shell commands to open files.
Sanitization: The instructions do not specify sanitization for codebase content interpolated into the HTML report. The Mermaid.js configuration in the report uses securityLevel: "loose", which increases the potential impact if malicious diagram syntax is injected from the codebase.

improve-codebase-architecture