The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches font assets from Google's official GitHub repository to handle brand styling, which is a trusted source.
[COMMAND_EXECUTION]: Orchestrates video editing tasks by calling binaries such as ffmpeg, ffprobe, and whisper-cli using the subprocess module. These calls are implemented using list-based arguments without a shell wrapper, which effectively prevents command injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted media files and transcription data, which presents a surface for indirect prompt injection attacks.
Ingestion points: User-supplied video files and JSON-formatted word timings are ingested in the build_reel.py script.
Boundary markers: The workflow documentation defines manual steps for transcription cleaning and cut verification to maintain the integrity of the output.
Capability inventory: The skill has extensive capabilities for file system read/write operations and subprocess execution across its script suite to perform video rendering.
Sanitization: Input parameters from the configuration spec are processed using standard library path functions and passed to subprocesses as discrete list arguments, providing inherent sanitization.
[SAFE]: The skill follows security best practices for secret management by instructing users to store the OPENAI_API_KEY in a .env file at the repository root instead of hardcoding credentials.

reel-recut