The Agent Skills Directory

[SAFE]: The skill instructions and reference materials focus entirely on pedagogical frameworks, such as Cognitive Load Theory, Mayer's Principles of Multimedia Learning, and Gagné's 9 Events of Instruction. All guidance is aligned with academic and professional best practices.
[EXTERNAL_DOWNLOADS]: The skill recommends several well-known and trusted external services for design assets and accessibility validation, including Unsplash, Wikimedia Commons, The Noun Project, and WebAIM. These references are documented neutrally as standard resources for educational design and do not involve risky or unknown sources.
[INDIRECT_PROMPT_INJECTION]: The skill involves processing user-supplied educational content. (1) Ingestion points: User-provided learning objectives and domain content. (2) Boundary markers: The skill instructs the agent to structure data specifically within Gagné's 9 Events framework, which acts as a logical delimiter. (3) Capability inventory: The skill references external tool usage for actual file generation (e.g., a pptx skill) but contains no internal executable code. (4) Sanitization: No explicit data sanitization is mentioned, but the focus is purely on structural transformation. The risk surface is assessed as safe.
[NO_CODE]: The skill consists of markdown-based pedagogical instructions, structural templates, and validation checklists. It does not include embedded scripts, executable code, or load-time shell commands.

educational-presentation