Turn any video link into a file, frames, and a real transcript.
You have a video link — YouTube, TikTok, Instagram, or hundreds of other sites — and need the actual file, just a clip, just the audio, or a working transcript. Doing that by hand means learning yt-dlp's hundreds of flags and ffmpeg on top.
Give an AI coding assistant a real way to work with video: download from YouTube, TikTok, Instagram, and hundreds of other sites (full video, audio-only, subtitles, or just a clip), then break any video into what an AI can actually read — extracted frames, subtitle blocks, and metadata. Two paired skills, use one or both: - Video Download: wraps yt-dlp's hundreds of flags into plain requests ("just the audio," "just the clip between X and Y"). Verifies every download actually landed on disk before reporting success — never a false "done." - Video Analysis: gives the AI "eyes" on a video it can't play — samples frames, extracts burned-in subtitle text into a full transcript, and probes metadata (duration, resolution, audio, subtitle tracks) before deciding how to process it. Both self-heal on a new machine: they check their tool dependencies are actually working live, every time, instead of trusting a stale "already set up" flag — so they fix themselves silently instead of failing on an outdated assumption. Honest about limits: never claims to have "watched" a video (it reads extracted frames), and clearly distinguishes a subtitle-based transcript from an audio-based one.
Get this agent to leave a review — reviews are limited to people who actually used it.