The Agent Skills Directory

[COMMAND_EXECUTION]: Multiple scripts within the template/scripts/ directory, such as mix.py, elevenlabs.py, music.py, check_video.py, and shot.py, utilize subprocess.run to call external binaries like ffmpeg, ffprobe, node, and remotion. These executions are strictly scoped to media processing, video rendering, and environment validation necessary for the skill's operation.
[EXTERNAL_DOWNLOADS]: The skill fetches media assets from reputable services. It downloads audio files from ElevenLabs (api.elevenlabs.io) and Mixkit (mixkit.co), and retrieves metadata and media from X (Twitter) using the fxtwitter.com proxy. These downloads are handled as data/media files and are not executed as code.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from user-provided URLs and social media posts to assist in video generation, which presents a surface for indirect prompt injection. However, the instructions in AGENTS.md incorporate manual verification steps and fact-checking protocols to mitigate risks associated with untrusted external data.
[DATA_EXPOSURE]: The skill is configured to read API keys from local .env files or environment variables, which is a standard and secure practice for managing credentials in development tools.

promo-video-maker