The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides automated shell instructions for environment setup and document hosting. This includes using git, mkdir, cp, and chmod to manage files within the local repository and python3 -m http.server to launch a preview server bound to the local loopback address (127.0.0.1). These operations are scoped to the project directory.
[EXTERNAL_DOWNLOADS]: The documentation includes a reference command for manually fetching font assets from Google's official font registry (fonts.googleapis.com) to support self-hosting and offline rendering. This is documented as a maintenance task for a well-known service.
[PROMPT_INJECTION]: The skill processes user-supplied Markdown documents as input for document generation and a follow-up question-and-answer phase. This creates a surface for indirect prompt injection, where instructions embedded in a plan could attempt to influence the agent's behavior during the resolution of open questions or the accretion pass. The skill contains specific guidelines for the agent to maintain the authoritative status of the plan and verify output.
[SAFE]: The skill implements protective measures for its dynamic interactive features. The 'JS Playground' functionality is explicitly restricted to dependency-free, vanilla JavaScript with no network access, minimizing the risk associated with generated executable content.

qstack-plan-to-html