The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on extensive use of system commands to perform its tasks. It invokes node to execute dynamically generated JavaScript code (generate_slides.js) for PowerPoint creation. Additionally, it uses Python's subprocess module to call external utilities like soffice for PDF conversion, pdftoppm for image processing, pandoc for document format conversion, and xelatex for formula rendering.
[EXTERNAL_DOWNLOADS]: The skill instructions direct the agent to install several external dependencies via npm, pip, and brew (e.g., pptxgenjs, Pillow, Graphviz, mermaid-cli). While these are standard tools from reputable registries, the installation of third-party packages at runtime is a notable capability.
[PROMPT_INJECTION]: The skill features a 'Material Analysis' phase where it reads and processes user-provided academic papers or research notes. This serves as an ingestion point for untrusted data. Without explicit sanitization or strict boundary markers, there is a risk of indirect prompt injection where instructions hidden within a paper could influence the agent's behavior during the slide generation workflow.

powerpoint-slides