The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase data and interpolates it into a generated HTML report, creating a surface for indirect prompt injection.\n
Ingestion points: The skill reads git log --oneline, CONTEXT.md, and architectural decision records (ADRs) in docs/adr/, and uses a subagent to explore general codebase files (SKILL.md).\n
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the report generation process.\n
Capability inventory: The skill writes a file to the OS temp directory and executes shell commands (xdg-open, open, start) to launch the user's web browser (SKILL.md).\n
Sanitization: No sanitization or escaping of external content is defined. The report scaffold uses securityLevel: 'loose' for the Mermaid diagramming library, which allows HTML rendering and increases the risk of cross-site scripting (XSS) if malicious content is found in the scanned files (HTML-REPORT.md).\n- [EXTERNAL_DOWNLOADS]: The generated HTML report fetches external assets from well-known services at runtime.\n
Evidence: The report scaffold in HTML-REPORT.md references Tailwind CSS (cdn.tailwindcss.com) and the Mermaid library (cdn.jsdelivr.net) via CDN links.\n- [COMMAND_EXECUTION]: The skill utilizes standard shell commands to analyze the codebase and present results.\n
Evidence: SKILL.md uses git log to identify file hot spots and platform-specific commands like xdg-open or start to display the generated HTML report.

improve-codebase-architecture