The Agent Skills Directory

[SAFE]: The skill is a well-documented utility for media management. It contains no obfuscation, credential exfiltration, or malicious prompt injection patterns. Its operational directives include safety guidelines, such as refusing to assist with DRM circumvention.
[COMMAND_EXECUTION]: The provided Python scripts in SKILL.md and gpt.md use the subprocess module to execute ffmpeg and ffprobe. These implementations correctly use list-based arguments for command execution, which mitigates the risk of shell injection from untrusted filenames.
[DATA_EXPOSURE]: The skill instructions and scripts access local media files and metadata (via ffprobe) for the intended purpose of trimming and joining video segments. No sensitive system files or credentials are targeted.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests file paths provided by the user, it does not interpret data from those files as instructions, and its use of structured subprocess calls prevents the file metadata from influencing the underlying system execution environment.

losslesscut