The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of uploaded slide decks and speaker notes, which could contain instructions intended to override agent behavior.
Ingestion points: Processes user-uploaded PPT files and associated speaker notes (SKILL.md).
Boundary markers: Absent; there are no specific instructions to the model to ignore potential commands embedded within the slide content.
Capability inventory: Script rewriting, narration generation, and video scene editing via Clueso MCP.
Sanitization: Absent; the skill does not specify validation or filtering of the slide content before processing.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for the user to manually connect remote MCP servers and plugins from external sources.
Evidence: Instructions include claude mcp add --transport http Clueso https://connect.clueso.io/mcp and codex plugin add app-6a310833394081919705b0a3ee745267@openai-curated-remote.
Note: These resources originate from the vendor's domain (clueso.io) or platform-curated registries.

slides-to-video