The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to save generated HTML files to specific filesystem paths, such as /tmp/html/[slug].html or within a project's public/ directory, and then execute the open <path> command to display the document to the user.
[INDIRECT_PROMPT_INJECTION]: The skill is intended to process and render external, potentially untrusted data (e.g., code reviews, PR descriptions, and transcripts) into HTML artifacts.
Ingestion points: The skill ingests user prompts, code snippets, PR bodies, and meeting transcripts as described in SKILL.md and references/inspiracion.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands within the processed data are defined.
Capability inventory: The agent performs file writes to /tmp and public/ directories and executes the open command.
Sanitization: There is no explicit requirement or instruction for sanitizing the external data before it is interpolated into the HTML output.
[DYNAMIC_EXECUTION]: The skill generates functional HTML documents at runtime, including embedded CSS and simple JavaScript for features like sliders, drag-and-drop interactions, and clipboard management.
[EXTERNAL_DOWNLOADS]: The skill references font assets from Google's official font service (fonts.googleapis.com) to maintain visual identity.

html-perennia