The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions in SKILL.md are focused on architectural principles and software modeling. There are no patterns suggesting attempts to bypass safety filters or override core agent instructions. Natural instructional language like 'IMPORTANT' is used strictly for pedagogical purposes.
[DATA_EXFILTRATION]: No network-capable commands (e.g., curl, wget, requests) or hardcoded credentials were found across the scripts and markdown files. The skill does not access sensitive system paths.
[REMOTE_CODE_EXECUTION]: The skill does not download or execute scripts from remote sources. The scaffold.py script is a local utility using standard Python libraries for file templating.
[COMMAND_EXECUTION]: The Python script scripts/scaffold.py uses standard filesystem operations (pathlib) to generate code templates. While it writes files to disk, it is a benign developer tool intended for local boilerplate generation and is not invoked by the agent instructions themselves.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided code for review. While this constitutes an ingestion surface for untrusted data, the agent lacks tools to execute that code or perform network operations, mitigating the risk to text-based manipulation only. No boundary markers are explicitly defined in the instructions, but the scope of the agent's capabilities is limited to architectural advice.

domain-driven-design