Claude motion faceless skills are a Claude Code skill set that renders 74 animated explainer styles, including whiteboard, chalkboard, claymation, CRT terminal, Game Boy and manga, frame by frame in headless Chrome and writes them to MP4 at 16:9 or 9:16. Each style ships as original code with a canvas renderer and a synthesized sound score. There are no paid APIs and no generation credits. It is Kevin’s open-source project at github.com/kevinbadi/claude-motion-faceless-skills.
your prompt in Claude Code
|
v
skill picks a style (1 of 74)
|
v
canvas renderer draws --> frame by frame
synthesized sound score in headless Chrome
|
v
ffmpeg encodes --> out.mp4 (16:9 or 9:16)
|
v
you review, tweak the prompt, re-render
What Are Claude Motion Faceless Skills?
They are a Claude Code skill that turns a written prompt into a finished animated explainer with no face on camera.
Faceless channels live or die on a consistent look. The hard part is not the idea, it is producing the same visual language every week without a motion designer or a subscription render farm. This project takes that job and puts it inside Claude Code, the terminal agent from Anthropic.
Two ideas are bundled together:
- The styles. 74 named looks. Whiteboard, chalkboard, claymation, CRT terminal, Game Boy, manga, ukiyo-e, blueprint, synthwave and more. You pick a look once and reuse it across a series.
- The renderer. Every style is original code, not a template you license. A canvas renderer draws each frame and a synthesized sound score plays underneath. Headless Chrome renders the frames, then ffmpeg encodes them to MP4.
That means the output is deterministic. The same prompt and the same style give you the same video, which is what you want when a series needs to look identical for 40 episodes. Kevin walks through builds like this on the KevBuildsApps YouTube channel.
How the 74 Faceless Explainer Styles Work (No Paid APIs, No Credits)
Nothing in the pipeline calls a hosted generation service, so there is no per-video credit to buy and no rate limit to hit.
The README is explicit: every style is original code that draws to a canvas and writes frames through headless Chrome. The audio is synthesized, not sampled from a library. Your machine does the work.
| Part | What it does | What it needs |
|---|---|---|
| Style | Defines the look: typography, palette, motion, sound | Chosen from the 74 included styles |
| Canvas renderer | Draws each frame of the animation | Node 18 or later |
| Sound score | Synthesized audio track per style | Runs locally with the render |
| Headless Chrome | Paints frames off screen | Google Chrome installed |
| ffmpeg | Encodes frames into an MP4 | ffmpeg on your PATH |
The practical upshot: a 30 second vertical explainer costs you render time, not credits. Once the toolchain is installed you can iterate on a hook ten times in an afternoon and keep the version you like.
How to Install the Claude Code Skill Locally
Install takes one clone, one script and four prerequisites, all of which the README names.
Before you start, confirm you have Claude Code, Node 18 or later, Google Chrome, and ffmpeg on your PATH.
git clone https://github.com/kevinbadi/claude-motion-faceless-skills.git
cd claude-motion-faceless-skills
./install.shThat command copies the skill into ~/.claude/skills and installs the renderer, so every Claude Code session on the machine can use it.
If you only want the skill available inside one repository, point the installer at that project:
./install.sh --project /path/to/your/projectThat copies into .claude/skills/ instead. On macOS, if Chrome is not in the default location, set the CHROME_PATH environment variable before you render so headless Chrome can find the binary.
The source lives at github.com/kevinbadi/claude-motion-faceless-skills. Read the README before you file an issue; it is the authority on flags and requirements.
How to Use It in Claude Code: Prompts That Render MP4
You drive it in plain language. Claude Code reads the skill, picks a style, and runs the render.
The README gives two example prompts. Start with a whiteboard explainer:
Make a 30 second vertical whiteboard explainer on how AI
agents can run a faceless YouTube channel.Then ask to see alternatives before you commit:
Show me the claymation and CRT terminal versions of that same
script, 9:16, so I can pick one for the series.A workable loop for a real channel looks like this:
- Write the 3 to 5 line script yourself. The style does the visuals, not the argument.
- Ask for the same script in two or three styles at 9:16.
- Watch all of them, keep the one that survives a phone screen.
- Lock that style and re-render the rest of the series in it.
Because the render is local, feedback is a re-run, not a new purchase. The launch video at youtu.be/-QBJH_PK3pY shows the kind of output this approach produces.
How to Render Vertical Shorts vs 16:9 Explainer Video
Aspect ratio is a render setting, and the two formats want different scripts, not just a different frame.
The project writes MP4 at 16:9 or 9:16. Same style, same renderer, different canvas. What changes is how you write for it.
| 9:16 vertical | 16:9 long form | |
|---|---|---|
| Where it lands | Shorts, TikTok, Reels | YouTube long form, embedded explainers |
| Script length | One idea, roughly 20 to 40 seconds | Three to six beats, several minutes |
| First frame | Has to read at arm’s length | Can carry a title card and context |
| Text density | Few words, large type | More room for labels and diagrams |
| Best styles | Whiteboard, chalkboard, Game Boy, CRT terminal | Blueprint, synthwave, ukiyo-e, manga panels |
| Series habit | Same style every upload so the feed recognises you | Same style per playlist, looser across the channel |
A useful test: render the same 30 second script in both ratios. If the vertical version feels crowded, the script is too long, not the frame too small.
Which Styles to Pick for a Faceless YouTube Channel
Pick one style per series and treat it like a brand asset, because switching looks monthly resets your recognition.
The 74 styles fall into recognisable families. Match the family to the subject, then commit:
- Teaching a process: whiteboard and chalkboard. Hand-drawn marks read as explanation, so tutorials and breakdowns fit.
- Tech and AI commentary: CRT terminal and Game Boy. Screen-era looks signal “this is about software” before anyone hears a word.
- Story and history: ukiyo-e and manga. Strong visual identity, good for narrative channels that need mood.
- Design, systems and planning: blueprint. Clean lines suit architecture, workflows and diagrams.
- Music, retro and culture: synthwave. Instant genre signal.
- Playful explainers: claymation. Soft, tactile and hard to confuse with a stock slide deck.
One operational note. Faceless does not mean scriptless. The style carries attention for the first few seconds; the script has to carry the rest. Write the claim first, then choose the look that makes the claim land.
Claude Skills for Marketing: Where This Fits Your Workflow
The render is one step. The rest of a marketing workflow is scripting, publishing and reading what happened.
The usual flow for a faceless channel is: idea, script, render, caption, publish, then read the numbers and change the next video. This project covers the render. You still need a place to publish to your channels and a way to see how each video did.
That is where Creator OS fits, through facts rather than overlap. Creator OS is one place to publish, schedule, reply and read analytics on 11 platforms plus Skool. It ships a hosted MCP server and a CLI, so an AI agent can run your accounts. A Claude Code session can render the MP4 with this skill, then hand the file to Creator OS through the MCP server or the CLI to post it.
Two details worth knowing. The CLI has no Reddit, Discord, Slack, WhatsApp or Shopify commands; those platforms are reached through the API, the MCP server or posts:create. And Creator OS reads watch stats for Instagram Reels, including skip rate in the first 3 seconds and average watch time, so an agent can compare videos and adjust the next hook.
For related setup and workflow posts, see the Claude Code and Creator OS MCP setup guide, the one line launch video skill, and how to build a social media agent with Claude Code.
FAQ
What are Claude motion faceless skills?
They are a Claude Code skill set that renders 74 animated explainer styles, such as whiteboard, claymation, CRT terminal and manga, into MP4 video without anyone appearing on camera. Each style is original code: a canvas renderer plus a synthesized sound score, drawn frame by frame in headless Chrome and encoded with ffmpeg. It is Kevin’s open-source project.
Can I use Claude Code to make faceless YouTube videos for free?
The project itself has no paid APIs and no generation credits, and the README describes no usage fee for the skill. You still need Claude Code, Node 18 or later, Google Chrome and ffmpeg installed locally, and your own machine does the rendering. Check the repository for the current license and terms before you build a channel on it.
Do Claude Motion styles need paid APIs or generation credits?
No. Every style is original code that draws to a canvas and renders through headless Chrome, and the audio is synthesized rather than licensed. There is nothing to top up and no per-render meter. The cost is your hardware and time.
How do I install Claude skills from Claude Code?
For this project, clone the repository and run ./install.sh, which copies the skill into ~/.claude/skills and installs the renderer. Add --project /path/to/your/project to install it into a single project’s .claude/skills/ folder instead. For other skills, check the source project’s own instructions, since install steps differ.
What video styles can I render with Claude Motion?
The README lists 74 styles, naming whiteboard, chalkboard, claymation, CRT terminal, Game Boy, manga, ukiyo-e, blueprint and synthwave among them. Each can render at 16:9 or 9:16, so the same style works for long form and for vertical Shorts. Read the repository README for the full list.
Get started
Start with the skill. Clone claude-motion-faceless-skills, run the installer, and render one 30 second vertical explainer in a style you could live with for a year. Watch the walkthroughs on the KevBuildsApps YouTube channel if you want to see the loop in motion.
Then put a publishing layer behind it. Creator OS plans start at $14.99/month for 11 platforms plus Skool, with the API, the MCP server and the CLI included, so the same agent that rendered the MP4 can schedule it and read the analytics afterward. Create an account at creatoros.ca/sign-up and connect your channels. The docs at creatoros.ca/docs cover the MCP setup, the CLI and the post options.
Get started for $0
Free trial for new subscribers. Cancel anytime.