The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes well-known and official Node.js packages puppeteer and cheerio to handle HTML parsing and image rendering, as specified in package.json.
[COMMAND_EXECUTION]: The skill involves the agent executing a local Node.js script, generate-images.js, to process generated HTML files and capture screenshots using a headless browser. This behavior is documented and central to the skill's primary function of image generation.
[PROMPT_INJECTION]: The skill processes untrusted Markdown content provided by users to generate formatted HTML output, which represents a standard indirect prompt injection surface.
Ingestion points: Markdown article content or file paths provided in the agent prompt (SKILL.md, README.md).
Boundary markers: The skill does not employ specific delimiters or instructions to the agent to disregard formatting-related instructions within the source text.
Capability inventory: The skill performs local filesystem writes to the output/ directory and executes a headless browser instance via Puppeteer.
Sanitization: Content is rendered in Puppeteer without an explicit HTML sanitization layer, though it is executed within a headless browser context.

mj-format