The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads standard NPM packages and browser binaries via Playwright during its documented setup process.
[COMMAND_EXECUTION]: The tool operates through the execution of Node.js scripts and CLI commands provided by its dependencies.
[DATA_EXFILTRATION]: The secondary script build_pdf_ofm.mjs contains hardcoded absolute file paths pointing to a specific user's directory on a Windows system (C:/Users/mphan/...), which exposes the author's local directory structure but does not impact user data security.
[INDIRECT_PROMPT_INJECTION]: The skill processes and renders Markdown/HTML input in a browser environment, which is an inherent capability of the tool. Untrusted input could theoretically include scripts or external resource requests.
Ingestion points: scripts/render.mjs reads content from a file path supplied via the --in command-line argument.
Boundary markers: No explicit delimiters or security warnings are used to isolate the rendered content from the execution environment.
Capability inventory: The tool uses Playwright (Chromium) to render content, allowing for complex web interactions including JavaScript execution and network requests.
Sanitization: No sanitization is performed on the input content, as the primary goal is full fidelity rendering of the provided markup.

mph-figure