The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill instructs the agent to create a Node.js script (presentation/generate.js) based on patterns in references/slide-patterns.md and then execute that script to produce a .pptx file. This involves runtime code generation and execution.
[COMMAND_EXECUTION]: The workflow requires executing several shell commands, including npm install, running the generated node script, and using python -m markitdown to verify the output.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from multiple sources without explicit sanitization or boundary markers.
Ingestion points: Reads content from docs/course-description.md, docs/chapters/*/index.md, and files within input-knowledge/.
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the textbook files.
Capability inventory: The agent has the capability to write files (generate.js), execute shell commands (node, python), and read project files.
Sanitization: No escaping or validation of the external textbook content is mentioned before it is used to generate the script or slide content.

textbook-to-presentation-generator