The Agent Skills Directory

[SAFE]: The skill implements standard translation workflows for nonfiction content. It includes clear documentation for preferences management via EXTEND.md and correctly handles local file system operations for creating output directories and temporary analysis files.
[COMMAND_EXECUTION]: The skill uses shell commands for environment checks (e.g., checking for bun or codex versions) and file existence checks. These are used for operational logic and do not involve unsanitized user input or privilege escalation.
[EXTERNAL_DOWNLOADS]: The skill references standard package registries (npm) for its dependencies (markdown-it) and uses the BUN_X runtime. These are well-known developer tools and do not represent a security risk.
[REMOTE_CODE_EXECUTION]: The optional Codex engine path (--engine codex) uses codex exec with explicit sandboxing (read-only and workspace-write). This is a documented feature of the specific platform integration intended for model-assisted tasks.
[DATA_EXFILTRATION]: No exfiltration patterns were detected. Network access is mentioned for fetching content if a URL is provided as a source, which is a core functional requirement of the translation skill.

editorial-translation