The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/shoot.mjs file uses node:child_process.spawn to launch a headless Google Chrome instance from a hardcoded macOS path. This is used to automate the rendering and screenshotting of UI components.
[DYNAMIC_EXECUTION]: The scripts/shoot.mjs script utilizes the Chrome DevTools Protocol (Runtime.evaluate) to execute JavaScript strings within the browser context. This allows the skill to manipulate the DOM, trigger React re-renders, and extract visual metadata from the page.
[EXTERNAL_DOWNLOADS]: The scripts/shoot.mjs script requires the ws Node.js package for WebSocket communication with the browser's debugging port. This introduces an external dependency not found in the standard library.
[INDIRECT_PROMPT_INJECTION]: The skill provides a mechanism for processing user-supplied SVG content through scripts/central.py.
Ingestion points: SVG data is read from stdin in scripts/central.py.
Boundary markers: The script does not use explicit boundary markers or instructions to ignore embedded content.
Capability inventory: The skill includes scripts with process spawning (spawn) and file system access (mkdirSync, writeFileSync).
Sanitization: scripts/central.py uses regular expressions to strip specific SVG attributes, but it does not perform full schema validation of the untrusted XML/SVG input.

add-icon