The Agent Skills Directory

[COMMAND_EXECUTION]: The script process_video.py orchestrates a multi-step workflow by invoking external Python scripts (such as transcribe_video.py and suggest_chapters.py) and the system's ffmpeg utility using the subprocess.run() function. Arguments are passed as lists, which is a secure practice that prevents shell injection. File paths are managed through the pathlib library to ensure safe and platform-independent resolution.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data (video content) through a transcription service, the output of which is subsequently used by an AI model to generate chapter suggestions.\n
Ingestion points: Video file content is transcribed into transcript.json by process_video.py.\n
Boundary markers: The skill does not implement explicit delimiters or instructions to the AI model to ignore or separate potential instructions embedded within the transcript data.\n
Capability inventory: The skill has permissions to read and write files, and can execute shell commands via the Bash tool and subprocess.run() calls (specifically executing local scripts and ffmpeg).\n
Sanitization: No sanitization, validation, or filtering of the transcribed text is performed before it is provided as input to the chapter suggestion logic.

video-full-process