The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local repository, which could be manipulated to influence the agent's architectural recommendations or behavior.
Ingestion points: Commit history (git log), domain glossary (CONTEXT.md), architectural records (docs/adr/), and various source code files during organic exploration.
Boundary markers: The instructions do not specify any delimiters or safety prompts to prevent the sub-agent from following instructions embedded within the scanned codebase files.
Capability inventory: The skill can execute shell commands, write files to temporary directories, modify local documentation (CONTEXT.md), and call other agent skills for domain modeling.
Sanitization: There is no explicit sanitization or filtering of codebase content before it is processed by the agent or rendered into the HTML report.
[COMMAND_EXECUTION]: The skill uses shell commands to inspect the repository environment and present findings to the user.
Evidence: Executes git log --oneline to identify codebase activity hotspots.
Evidence: Uses platform-specific commands (xdg-open, open, start) to automatically open the generated HTML report from the temporary directory.
[EXTERNAL_DOWNLOADS]: The generated architectural report includes references to external assets from well-known content delivery networks.
Evidence: The HTML template in HTML-REPORT.md loads the Tailwind CSS framework from cdn.tailwindcss.com.
Evidence: The HTML template imports the Mermaid diagramming library from cdn.jsdelivr.net.

improve-codebase-architecture