The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Python script (scripts/extract-pptx.py) to process user-provided PowerPoint files. This operation is limited to extracting text and images to a local directory.
[EXTERNAL_DOWNLOADS]: The skill references standard Python libraries (python-pptx, Pillow) and well-known external font providers (Google Fonts and Fontshare) required for its core presentation generation features.
[DATA_EXPOSURE]: Processes local PowerPoint files and creates asset folders in the project directory. There is no evidence of unauthorized sensitive file access or network-based data exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from PowerPoint files and user notes.
Ingestion points: .pptx files and user-provided text (SKILL.md Phase 4 and Phase 1).
Boundary markers: Content is wrapped in HTML tags within the template (html-template.md) but lacks explicit injection delimiters.
Capability inventory: File-system writing, Python script execution, and browser opening (SKILL.md).
Sanitization: No explicit sanitization of extracted PPT text before interpolation into the HTML template, though the data is rendered in a sandboxed browser environment.

frontend-slides