The Agent Skills Directory

[SAFE]: The skill instructions provide legitimate logic for simplifying codebases, including rules for inlining short, single-use functions and consolidating thin types. These operations are intended to reduce cognitive load and do not involve dangerous command execution or system modification.
[SAFE]: The skill directs the agent to read project configuration files like CLAUDE.md to ensure compliance with local coding standards. This is a common and safe practice for development tools and does not constitute a data exposure risk as the files are project-specific and not sensitive system credentials.

ddt