The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes system utilities ffmpeg and ffprobe via child_process.spawnSync and execFileSync to process media and encode video frames. This is a core functional requirement for the skill's operation and is located in categories/maps/bake-basemap.mjs and grounding/locate.mjs.
[EXTERNAL_DOWNLOADS]: The skill downloads font files, map data, and JavaScript libraries from well-known and trusted sources, including Google APIs (googleapis.com), JSDelivr (cdn.jsdelivr.net), and official map tile providers such as Esri (arcgisonline.com) and CARTO (cartocdn.com). It also uses Puppeteer to render content from these sources.
[REMOTE_CODE_EXECUTION]: The skill utilizes npx to execute components of the hyperframes ecosystem for project initialization, linting, and rendering as described in the SKILL.md file. These commands fetch and execute packages from the NPM registry.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from sources such as web pages, news articles, and tweets (categories/webpage/module.md, categories/news/module.md), which represents a potential attack surface.
Ingestion points: External content is ingested via hyperframes capture and automated search tools.
Boundary markers: The instructions provide specific content extraction rules, which help ground the model, though they do not include explicit adversarial filtering.
Capability inventory: The skill possesses capabilities for command execution (ffmpeg), headless browser execution (puppeteer), and network operations.
Sanitization: The documentation notes SVG sanitization for specific asset imports (Figma), but no global sanitization for other raw text content.

motion-graphics