The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the artifacts tool to execute generated JavaScript for slide creation and manipulation as its primary function.
[DATA_EXFILTRATION]: Documentation in SKILL.md and references/presentation.md encourages using node:fs/promises to read local files and fetching remote images. This combination of local file system access and network capabilities creates a vector for data exfiltration if the agent is manipulated into reading sensitive data and transmitting it externally.
[PROMPT_INJECTION]: The ability to import external files via PresentationFile.importPptx presents a surface for indirect prompt injection from potentially malicious PowerPoint documents.
Ingestion points: PresentationFile.importPptx in references/presentation.md allows loading existing .pptx files into the session context.
Boundary markers: None identified; the instructions do not specify any markers to isolate or prioritize the skill's logic over content within imported files.
Capability inventory: JavaScript execution via the artifacts tool, file system read/write access via node:fs, and network fetching capabilities.
Sanitization: The skill provides no guidance or mechanisms for sanitizing or validating the data or instructions contained within imported slides.

slides