The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several system utilities to process media and manage files.
It calls ffmpeg and ffprobe to perform frame extraction and metadata analysis on user-supplied videos.
It utilizes whisper for transcribing audio content to derive dialogue scripts.
It uses the open command on macOS to allow the user to quickly review output directories.
[PROMPT_INJECTION]: The skill is designed to process untrusted external media, creating a surface for indirect prompt injection.
Ingestion points: Video and audio files provided by the user for the analyze-video and clone-ad workflows.
Boundary markers: The skill strictly enforces a "dialogue confirmation gate," requiring the agent to extract and present the script for explicit human approval before any generation occurs.
Capability inventory: The environment utilizes ffmpeg, whisper, and curl for its core operations.
Sanitization: Employs manual review checkpoints and shell quoting in utility scripts to prevent malicious payload execution.
[EXTERNAL_DOWNLOADS]: The skill communicates with external-api.arcads.ai to perform its primary function.
This domain is the official endpoint for the Arcads service.
The skill instructs the user to store credentials in a .env file to prevent accidental exposure in logs or chat history.

arcads-external-api