The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local command-line tools including ffmpeg, ffprobe, and npx hyperframes to process video assets and perform rendering operations. These operations are restricted to the project workspace and are functional requirements for media production.
[EXTERNAL_DOWNLOADS]: The skill fetches libraries and data from well-known content delivery networks, specifically cdn.jsdelivr.net for MapLibre and TopoJSON, and arcgisonline.com for geographic tiles. These resources are from established service providers and are used according to standard development practices.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface (Category 8) when handling data from external URLs, news articles, or social media posts.
Ingestion points: Data is ingested from user-provided URLs and external search results during the sourcing phase.
Boundary markers: Prompts for subagents do not utilize explicit delimiters to segregate untrusted source data from system instructions.
Capability inventory: The agent can perform network requests and execute shell-based rendering commands.
Sanitization: There is no dedicated validation or sanitization layer for external content prior to its inclusion in generated HTML templates.

motion-graphics