The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified during the analysis of the skill's logic and scripts.
[COMMAND_EXECUTION]: The skill frequently executes ffmpeg and ffprobe to perform media assembly. These executions are performed using the subprocess module with argument lists and without the shell=True parameter. This architecture prevents shell injection attacks, even when processing file paths or metadata provided as input.
[INDIRECT_PROMPT_INJECTION]: The skill ingests several JSON files, such as tts_meta.json and visual_overlays.json, which contain text for subtitles and on-screen overlays. The skill correctly implements escaping logic (e.g., _escape_drawtext_text in visual_render.py) to ensure that this text cannot break out of FFmpeg filter syntax or influence the execution flow.
[DATA_EXPOSURE]: File operations are strictly scoped to user-defined working directories and explicitly provided media paths. The skill uses path resolution and strictly validates file existence and types, preventing unauthorized access to sensitive system files.

video-assemble