The Agent Skills Directory

[COMMAND_EXECUTION]: The Python scripts chapter_project.py and render_chapter_bar.py utilize subprocess.run to call ffmpeg and ffprobe. These calls use argument lists rather than shell strings, which is a secure method for calling external media tools required for the skill's primary purpose of video rendering and metadata extraction.
[EXTERNAL_DOWNLOADS]: The project relies on standard external dependencies including the Python Pillow library and several Node.js packages for the React-based studio component. These are installed through official package managers (pip and npm) as described in the documentation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from subtitle files (.srt, .vtt) to generate chapter plans. While this represents an ingestion surface for untrusted text, the scripts focus on extracting timestamps and titles for rendering, and provide basic sanitization by stripping HTML tags from the transcript content.
[DATA_EXPOSURE]: The React studio application operates locally in the user's browser, using URL.createObjectURL to preview local video files without uploading them to external servers. No unauthorized network operations or credential leaks were detected.

lov-video-chapter