The Agent Skills Directory

[SAFE]: The skill defines a standard operational workflow for media acquisition. It includes best practices such as quoting model costs before spending credits, adhering to license-safe audio sources (Lyria, CC0), and following a prioritized sourcing order (library, then stock, then generation). The tools referenced, such as media_index and generateMedia, are consistent with the vendor's intended functionality.
[INDIRECT_PROMPT_INJECTION]: The skill acknowledges a potential surface for indirect prompt injection when processing external or generated media. • Ingestion points: Untrusted data enters the agent context through the outputs of fetchStockMedia and generateMedia (SKILL.md). • Boundary markers: Absent from the workflow description. • Capability inventory: Includes media_index, fetchStockMedia, generateMedia, addHtmlElement, addShapeClip, and replaceClipMedia (SKILL.md). • Sanitization: While no technical sanitization is described, the skill provides a procedural safeguard by instructing the agent to visually verify content rather than trusting metadata.

media-sourcing