Adds ltx2-vidgen-skill to Creative & Media (alphabetical, external-link format with attribution).
What it solves: hosted AI-video APIs charge per clip and rate-limit you, which kills iteration. This is a Claude Code skill that deploys your own LTX-2.3 (22B) video backend to your Modal serverless-GPU account, then drives it from Claude Code — drop a photo (or a prompt), get a video. Your GPU, your bill, no per-clip meter.
Who uses it: creators making reels/shorts who want to explore a prompt many ways cheaply (--variations N) and keep the best take for a few cents.
Modes: text-to-video, image-to-video, keyframe interpolation, video-to-video (retake/restyle), and IC-LoRA canny/depth/pose control — with synced audio. Reel/YouTube/square format presets.
Example: in Claude Code — "turn this photo into a video, subtle natural motion, vertical reel" → saves the .mp4 + a preview frame.
Install: npx skills add patraxo/ltx2-vidgen-skill. Public, v1.0.0, CI-linted SKILL.md, demo in the README: https://github.com/patraxo/ltx2-vidgen-skill
Author submission. Single-line addition; external-repo entry matching the existing Creative & Media format (e.g. anydesign, imagen, Pixelbin).