About this skill
A character on the page watches the visitor. Move the pointer and its head turns to look; leave it alone and it glances about. Nothing is rendered live: the head is a set of frames made once, a grid of yaw and pitch, a strip, a look-around loop or a few held gazes, and a small player picks the pose nearest the pointer, crossfades toward it one step at a time and blinks across long jumps. The skill starts from a brief with the character, its rig and angle range, and its temperament as numbers: how fast it turns, how long it holds a pose, when it wanders. A renderer turns a three.js character into frames and blink sheets; a second script turns a look-around clip made with AdAnt video into frames by tracking one feature. The same skill builds a code-gated drop page with a self-typing headline, a full turn scrubbed by scroll with hotspots, and a film scrubbed by pointer height. A Playwright check measures coverage, pose error, jitter at rest, swap rate, memory, the idle wander, touch at 390 px, reduced motion and the page without JavaScript.
You say
Using this skill, ask me the brief's questions for my tea shop's launch page, render our clay owl mascot as a 9 by 5 pose grid that follows the cursor, add a drop code field it watches, then run the gaze check and show me the desktop and phone screenshots.
How it works
- Your agent asks the brief's questions: the character, where its frames come from, the rig and angle range, its temperament as numbers, and the page around it.
- It makes the frames, by rendering a 3D model or extracting them from a look-around clip, and checks a contact sheet for poses that pop.
- It builds the static page with a still pose first, then adds the gaze player and any drop gate or scroll turn the brief asks for.
- It runs the gaze check in Chrome, reads the report and the screenshots, fixes what failed, and records a preview clip.
Getting a good result
- Give the character a feeling about the visitor, then set the numbers from it: a sulky one turns slowly and holds a pose longer.
- Keep the light and the shadow fixed while rendering. Only the head should move from frame to frame.
- Choose the still pose as carefully as a hero image. Visitors with reduced motion see nothing else.
What you get
- A brief format with a field for every choice the model would otherwise guess, and a filled example
- A gaze player for grid, strip, loop and held-pose rigs, with crossfades, blinks, an idle wander, touch and keyboard focus
- A renderer that turns a three.js character into pose frames and blink sheets, and a script that turns a look-around clip into frames
- An example site: a fictional knitwear label whose felted hedgehog watches the pointer, guards a drop code and turns on scroll; one command renders its frames
- A Playwright check for coverage, pose error, jitter, swap rate, memory, wander, phones, reduced motion and no JavaScript, and a clip recorder
Good fit
- Launch, drop and waitlist pages that need a mascot people remember
- Brands with a character, toy or 3D model already, or a still to animate with AdAnt video
- Agents that write front-end code and can run Node and a browser
Not for
- Characters that must act out a scene or speak; this one only looks
- Pages where every kilobyte counts: the frames are 1 to 3 MB
- A real person's face without their written consent
How to install it
- Download the ZIP and unzip it.
- Move the folder holding
SKILL.mdinto your agent's skills folder, such as~/.claude/skills/for Claude Code. - Start a new session and ask for what you want to make.
Needs: An agent that reads skills, Node 20 or later and Playwright with Chrome, for the renderer, the check and the recorder, ffmpeg, for frames from a clip and the preview clip, AdAnt credits only if you generate the character's clip with AdAnt video.
Popular on X
Posts showing demand for this kind of video. Unless a post names this tool, it wasn't made with it.
- A post listing five UI and UX sites for builders, this library among them 98,508 views on X, measured 8 November 2026
FAQ
What does it cost?
$2.99 once for this skill, or $99 once for every skill in the library. It also comes with every paid AdAnt plan, from $15 a month. Rendering a 3D character, the check and the recorder all run on your own computer and use no AdAnt credits. Only generating a look-around clip with AdAnt video costs credits.
Can I use it in commercial work?
Yes. The skill's code is MIT licensed: use it and what it makes in your own commercial work, including for clients, and keep the license file with any copy you pass on.
Where do the character's frames come from?
Three ways. Render a 3D model with the bundled script: the example's 45 poses with their blink sheets, and its 36-frame turn, render in about 6 seconds on an Apple M-series laptop. Generate a clip of the character looking around with AdAnt video and extract frames with the second script, which tracks one feature to work out where each frame looks; on test clips with known angles it was off by 0.7 to 2.9 degrees at the median and 4.3 to 6.9 at the 95th percentile, depending on the feature tracked. Or film a puppet on a turntable.
How well does the example follow the pointer?
In headless Chrome on an Apple M-series laptop, the check found no gaze in the range further than 7.1 degrees from a pose, and the settled pose within 6.0 degrees of the pointer for 95% of 63 positions. A hand trembling on a pose boundary changed nothing and drew nothing, a fast scribble made at most 13 pose changes in a second, the 45 decoded frames took 55.4 MB, the idle wander began at exactly 6 seconds and gave the head back 16 ms after the pointer moved, and at 390 px wide a tap turned the head and the page did not scroll sideways.
Does it work with reduced motion, phones and Next.js?
Yes. With reduced motion the player draws one still pose and listens to nothing, and without JavaScript the same pose shows as an image. On phones the head follows the last touch. The skill's notes cover a client component that starts the player in an effect and destroys it on unmount, and keeping the pointer out of React state.
Is this the site that inspired it?
No. The idea comes from a library of scroll-driven website prompts, linked on this page, whose prompts are only for its members. AdAnt wrote this skill, its player, scripts and example character without them, and the library's makers did not write or review it. The clip on this page is AdAnt's example site, recorded with this skill's script.
Can I use what it makes commercially?
Yes. The pages and frames you make are yours. The scripts, player and example are MIT, and the license text is in the ZIP.
Related skills

Scroll-Scrubbed Film Site
Have your agent build a landing page where scrolling plays a film frame by frame, with chapters, callouts and a preloader, then check it in Chrome.

Dither & Glyph Hero
Have your agent build a hero that draws a live 3D scene in two inks, as dither, scanlines or glyphs, around an AI prompt, then check it in Chrome.

Drag Gallery Storefront
Have your agent build a shop you browse by dragging, with inertia, a seamless loop, a product sheet and a cart, then check it in Chrome.

Cinematic Video Hero
Have your agent build a one-screen hero from footage and type, with a freeze frame, text behind the subject and a loop, then measure it in a real browser.

Cursor Playground Hero
Have your agent build a hero section visitors play with, on small simulations that behave the same on every screen and never blow up.

Parallax Plates Site
Have your agent split one picture into depth plates and build a page that flies through them on scroll, then prove in Chrome that no plate edge ever shows.

Scroll-Driven Site
Have your agent build a landing page where scrolling pins, scrubs and reveals each scene, then measure it in a real browser.

Shader Reveal Site
Have your agent build a page where the pointer or the scroll uncovers a second picture through a shader, then measure the trail, the GPU and every fallback in Chrome.

Code-Rendered 3D Animated Short
Have your agent write a tiny wordless story, build its characters and set in three.js, shoot it from a shot list, score it and render a frame-exact MP4.

Particle Form Hero
Have your agent build a hero where GPU particles settle into your shape, morph to the next as the page scrolls, and part around the pointer.

Footage Landing Page
Have your agent build a full landing page under a looping footage hero, every word in one content file, then check it in Chrome.

Scroll-Posed 3D Product
Have your agent build a page where one 3D product holds the centre while the scroll poses it, section by section, until it docks into the buy box.