The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run in fetch_slides.py to execute the LibreOffice soffice binary for PPTX-to-PDF conversion. The execution uses a list of arguments and does not involve a shell, mitigating command injection risks.
[EXTERNAL_DOWNLOADS]: The skill makes network requests using httpx and playwright to download files or fetch presentation content from user-supplied URLs.
[INDIRECT_PROMPT_INJECTION]: The skill extracts text and links from remote RevealJS presentations and writes them to a markdown file for agent consumption, which is a common vulnerability surface.
Ingestion points: External HTML content from slide URLs processed in fetch_slides.py.
Boundary markers: None; extracted text is written directly without delimiters or warnings to the agent.
Capability inventory: File system writes, command execution via LibreOffice, and arbitrary network operations.
Sanitization: No sanitization or filtering of the extracted text content is performed before storage.

fetch-slides