The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run and subprocess.check_output to execute the ffprobe binary for metadata extraction. These commands are constructed using argument lists rather than shell strings, which effectively prevents shell injection vulnerabilities.
[DATA_EXPOSURE]: The scripts access local media files to retrieve metadata such as duration, resolution, and codec information. The access is restricted to file paths explicitly provided via command-line arguments, and the resulting data is returned as structured JSON to the agent without external exfiltration.
[REMOTE_CODE_EXECUTION]: All scripts are designed to run locally using the uv Python runner. They do not download or execute code from remote sources.

inspect