JSON2Video MCP

Hypit

An open-source agent workflow that takes a viral short video and rebuilds it as an editable template, letting you swap the presenter, script, voice and B-roll to produce many new variants with your own API keys.

html-video

Lets your coding agent turn HTML, CSS and data into real MP4 videos on your own machine, with templates and pluggable render engines.

explainroo

Lets your AI coding agent make narrated explainer videos: it writes the script, builds the visuals and renders the voice and video on your own computer, with no API key needed. Free and open source.

Vellm

Vellm turns photos of people and pets into short AI videos from ready-made templates. Pick one of 108 dance, funny, duo and pet templates built on ByteDance Seedance 2.5, upload your photos and it renders your own version, or start from a prompt in its Create Studio. A remote MCP server (https://api.vellm.app/mcp, OAuth sign-in) lets MCP-compatible AI apps create videos from your Vellm account.

LTX-2

Lightricks' open audio-and-video generation model. Official Python inference and LoRA trainer.

onetake

onetake is an agent skill for Claude Code and similar agents that makes product launch films, teasers and feature demos as one continuous camera move, where each beat grows out of the previous one instead of cutting between slides. The agent writes the motion as code and renders it with ffmpeg, and the skill checks continuity at every transition. The free edition on GitHub is for noncommercial use; onetake Pro adds a camera operator that follows each beat, source files for every example film, starter templates and a commercial licence.