The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text strings and media files, which presents a surface for indirect prompt injection if the resulting content or metadata is processed by an agent without sanitization.
Ingestion points: Untrusted text input and media file paths enter the context via command-line arguments in scripts/add_intro_outro.py, scripts/add_lower_third.py, scripts/add_subtitles.py, and scripts/add_watermark.py.
Boundary markers: Absent. There are no delimiters or warnings to ignore instructions that might be embedded in the text provided for subtitles or titles.
Capability inventory: The skill possesses file-writing capabilities using the write_videofile method in all core scripts (scripts/add_intro_outro.py, scripts/add_lower_third.py, scripts/add_subtitles.py, scripts/add_transition.py, and scripts/add_watermark.py).
Sanitization: Absent. User-provided strings are passed directly to MoviePy's TextClip objects without filtering, validation, or escaping.

video-overlay