The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses $ARGUMENTS to gather user requirements. This is standard input processing for planning and does not attempt to override agent safety guidelines.
[DATA_EXFILTRATION]: The skill has access to read the codebase and write documentation to the .claude/ directory. No network tools are requested or used, and no credentials are exposed.
[COMMAND_EXECUTION]: Uses file system tools (Read, Write, Edit, Grep, Glob) for legitimate purposes related to documentation and analysis. No arbitrary shell command execution or remote code downloads are present.

prp