The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates repository analysis by instructing the agent to execute shell commands such as git log and ls to identify system components and track change history for architectural evidence.
[COMMAND_EXECUTION]: The scripts/render_report.py utility executes system-level commands (e.g., open, start, xdg-open) via subprocess.run to display the generated HTML report in a web browser.
[SAFE]: The skill implements protective measures in its report rendering logic, including JSON encoding and escaping of HTML-sensitive sequences (</ to <\/) to prevent payload corruption or unintended script termination within the rendered report.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes architectural documents, ADRs, and source code from the target repository to inform its designs, though it implements structured verification and audit stages to maintain analysis integrity.

system-design