The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes local Python scripts (e.g., assets/check_path_naming.py and those in the examples/ directory) designed to validate project naming conventions and automate code transformations. These scripts are intended to be run locally by the developer via uv run.
[DATA_EXPOSURE]: Several example scripts access and process the local codebase to perform audits and refactoring (e.g., examples/bulk_refactor.py and examples/codebase_audit.py). This behavior is transparent and consistent with the skill's primary purpose of code maintenance.
[EXTERNAL_DOWNLOADS]: The documentation references standard and well-known Python packages (such as requests, numpy, and typer) and Node.js middleware for rate limiting. These are documented as architectural recommendations rather than hidden or suspicious downloads.
[PROMPT_INJECTION]: The skill provides structural guidance for code reviews and debugging (e.g., references/code-review.md), instructing the agent to maintain technical rigor and avoid performative agreement. These instructions are task-specific and do not attempt to bypass safety filters or extract system prompts.

python-ultimate