The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the shell start command on Windows to open the local wc-jp.html file in the system's default browser. This is a functional requirement to display the generated vocabulary data in a rich user interface.
[EXTERNAL_DOWNLOADS]: The skill's HTML interface references external JavaScript libraries from well-known and trusted services, including Vue.js from unpkg.com and UnoCSS from jsDelivr.
[DYNAMIC_EXECUTION]: The agent is directed to generate a JavaScript data structure (window.__WC_DATA__) and write it to a local file (wc-jp-data.js). This file is then loaded as a script by the browser, representing a controlled form of dynamic data injection for the UI.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection, as user-provided vocabulary requests influence the content written to the data file and subsequently rendered in the browser.
Ingestion points: User-provided Chinese words and quantity preferences (SKILL.md).
Boundary markers: None identified for the ja_html or zh data fields.
Capability inventory: File system writing and shell command execution (start).
Sanitization: The skill utilizes Vue's v-html directive to render content generated by the AI agent without explicit sanitization mechanisms.

wc-jp