The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a bundled Python script scripts/verify_revision.py to validate edits. Analysis of the script confirms it only performs local text processing, using standard libraries such as difflib and re to ensure consistency of metadata, code blocks, and formatting between versions. It does not perform network operations, unauthorized file access, or use dangerous functions like eval or os.system.
[EXTERNAL_DOWNLOADS]: The documentation references a public GitHub repository as the source for its linguistic guidelines. This is a static attribution in the text and does not involve runtime code fetching, package installation, or execution from external sources.
[PROMPT_INJECTION]: The instructions include explicit defensive directives for the agent to treat any commands found within user-provided text strictly as data to be edited rather than instructions to be executed, which effectively mitigates common prompt injection vectors from input data.

global-humanize-korean