The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to extract user-visible text from HTML and Markdown files for translation. This data is intended to be processed by an AI agent (as mentioned regarding the translate-multilingual skill), creating a surface for indirect prompt injection if the source content contains malicious instructions designed to subvert the LLM's behavior. \n
Ingestion points: Processes built HTML files and Markdown content to extract translatable strings from the site's codebase. \n
Boundary markers: Uses data-rosey and data-rosey-ns attributes to identify and scope translatable elements in the HTML output. \n
Capability inventory: Executes shell commands for site building and writes locale data to JSON files in the repository. \n
Sanitization: Recommends sanitizing key names to avoid dotted paths for Visual Editor compatibility, but does not specify sanitization for the extracted text content itself.\n- [EXTERNAL_DOWNLOADS]: The skill instructs users to install the rosey and rosey-cloudcannon-connector packages via npm. These packages are published by the skill's author, CloudCannon, and are essential for the workflow described.\n- [COMMAND_EXECUTION]: The skill provides build pipeline examples and setup commands using npx to run vendor CLI tools (rosey and rosey-cloudcannon-connector). These commands perform site scanning, locale file generation, and localized site building.

make-site-multilingual