givemotion: make videos with AI, free and open source

Ask your AI agent for a video and get a finished MP4. givemotion is a free, open-source (MIT) video engine that lets AI coding agents like Claude Code, Codex and Cursor make motion graphics for you: launch videos, explainers, animated charts, vertical shorts, UI demos and title sequences.

You describe the video in plain words. The agent builds it, looks at its own frames, fixes what is off, and renders the file. Then you ask for changes the way you would ask a motion designer, and only what you asked about changes.

It is free and open source (MIT): no account, no seats, no fee per video, zero npm dependencies, and you can fork it and make it your own. Created by Leonardo Grigorio as an independent project (formerly called motionwright). Full research paper and test videos: givemotion.org.

If you are looking for a way to make videos with Claude Code, an AI motion graphics tool you can run in your own project, or a free, open-source way to make videos with code (an MIT alternative to Remotion and HyperFrames), this is it.

Each of these was made by a fresh Claude Opus agent from a one-paragraph brief, using only this project's docs (data story and vertical short: 24 October 2026, round-3 remakes; lighthouse: 27 October 2026; givemotion 0.1.0 pre-release). The data story is in examples/data-story; more are in examples/ and on givemotion.org.

Try it in 1 minute

In your project folder, run:

npx givemotion init my-video

If you'd rather not use the terminal, ask your agent to do it: "Set up givemotion in this folder." Either way you get a starter video and the instructions your agent needs. Then ask for the video you want:

"Make a 20-second launch video for our app. Use our logo and colors, and put music under it."

The agent writes the video, runs the checks, looks at the frames, fixes what is wrong, and renders out/video.mp4. You can watch and scrub it live with npx givemotion dev my-video.

Then ask for changes the way you would ask a motion designer: "make the title blue", "hold the chart two seconds longer". The agent edits the video and renders again, and only the parts that changed are redone.

Already have a project? Add just the agent instructions:

# Installs .claude/skills/givemotion/ (Claude Code), .agents/skills/givemotion/ (Codex), and AGENTS.md (Cursor & others)
npx givemotion skill

# Or as a Claude Code plugin:
/plugin marketplace add givemotion/givemotion
/plugin install givemotion@givemotion

# Or with the Skills CLI:
npx skills add givemotion/givemotion

You need: Node.js 20+ and ffmpeg. Chrome is found on your machine or downloaded for you (npx givemotion install). Run npx givemotion doctor to check your setup and skill version.

How it differs from AI video generators

AI video generators (the models that turn a prompt into a clip) paint every frame from scratch. That is what you want for footage of things that were never filmed. It gets hard when you need one specific video: ask for one word to be fixed and the model makes the whole clip again, and something else comes out different. Text gets misspelled, logos get warped, and every attempt is paid for.

givemotion works the way a motion designer works. The agent builds the video out of real pieces (your logo file, your fonts, your exact words and numbers, shapes, illustrations, a camera, your music), and givemotion plays it back and records it frame by frame. For you that means:

• The same video every time. Render it again tomorrow and you get the same video. Nothing drifts between versions.
• Change one thing and only that changes. "Make the title blue" turns the title blue and leaves the rest as you approved it. Only the 1–2 second stretch of video the change touches is rendered again, byte-identical to a full render, so a small fix usually takes seconds.
• Text and logos come out right. Words are typeset, not drawn by a model, so they are sharp and spelled the way you wrote them. Your logo is your actual file.
• Cheap to make and to change. It renders on your own computer, with no fee per second of video. You pay for your AI agent as you already do; givemotion itself costs nothing to run.
• Many versions from one video. Swap the name, the numbers or the language (--data) and render as many copies as you need, while keeping the source file for next month's update.

What you can make: launch videos and app promos (with a cursor that clicks the real button), explainers with a camera that moves over a map or UI, hand-drawn whiteboard explainers (Excalidraw-style sketchy shapes and self-drawing arrows), live animated charts from your own numbers, vertical shorts for TikTok, Reels and Shorts with word-by-word captions, code explainers, logo stings, and transparent overlays (ProRes 4444 / WebM) for Premiere, DaVinci Resolve or Final Cut.

What it doesn't do is generate footage—it cannot film a photorealistic dog on a beach. If you need generated footage, make the clip with a video model and have the agent place it into a givemotion video.

Why it exists

I started givemotion because of how AI video models handle changes. When one detail is wrong, you ask again and the model makes the whole video again, and something else often comes out different. With code, a change is an edit: the agent changes one color or one timing and everything else stays as it was. I wanted an engine built around that loop—where the agent writes, checks, looks and fixes, and where a small fix costs a small render—and I wanted it under MIT so anyone, at any company size, can use it, fork it, or build a product on it without conditions.

It is an independent project, and it gets better in rounds: fresh agents make real videos from the docs alone, blind judges score them, and where the agents struggled becomes engine fixes.

— Leo

Why it works well with AI agents

Most video tools were built for people dragging layers on a timeline. AI agents work differently, so givemotion is built around how they actually behave:

1. It uses what AI already knows best: plain HTML, CSS and JavaScript. A video is one HTML file with no build step. Models have seen billions of web pages, so they write them well. In our October 2026 format test (givemotion 0.1.0 pre-release; 12 fresh Claude Opus agents, 4 briefs, 8 blind judges), plain HTML ranked first in 6 of 8 cases ahead of HTML with ready-made blocks and React (the experiment), matching outside research (DesignBench, 2025).
2. The agent can check its own work. An agent cannot watch a video in real time, so the engine does it for them:
   • check finds text that is cut off, off screen, covered, overlapping, too small or low in contrast, lines running through labels, two scenes' text smudged together in a transition, broken files and script errors, plus pacing faults (long dead stretches, an empty opening, text that leaves before you can read it, or a camera move that arrives too late), and says how to fix each one.
   • snapshot reads the timeline and picks the settled moments of each scene and camera shot into one contact sheet, so the agent sees the whole video at a glance instead of guessing timestamps.
   • review checks every timed beat of a brief (beats.png), builds motion strips at the hand-overs and busiest moments, and gives a scoring rubric for a fresh critic sub-agent. analyze does the same for any reference MP4 (--compare lists how yours differs).
3. Motion design tools are built in, not hand-coded from scratch. Camera moves and 3D worlds that blend like an operator, motion along paths, shape morphs, elements that fly from one scene into the next, 25+ transitions, kinetic type, particle bursts, live line charts the camera can ride, hand-drawn sketch shapes with line boil and tapered ink, labels pinned to shapes, rolling number readouts, and cuts that land on the music's beats are all built in.
4. Change one thing, render one part. Every 1–2 second segment carries a fingerprint of everything that can affect its pixels, and 1 worker and N workers produce the same bytes. After an edit, only the segments whose fingerprint changed are re-rendered.
5. Motion craft lives in the agent's docs. The skill pack has a short core (SKILL.md) plus topic files the agent reads only when needed: 15 render-verified technique recipes, 8 named looks, 9 genre guides, and 7 bundled OFL fonts (Inter, JetBrains Mono, Playfair Display, Space Grotesk, Bebas Neue, Nunito, Caveat) so renders look identical on every machine.

Tested with AI agents (research & evaluation)

Every test below ran with fresh, isolated agent sessions that saw only their own tool's official docs, scored by blind AI judges. The full paper, interactive video players, side-by-side frame crops and raw data are on givemotion.org and in [docs/experiments/](doc