Author’s ManimGX demonstration: a rainbow helicoid bends into a catenoid while preserving its intrinsic curvature.
Skill preview videoAbout this Skill
Give your agent a mathematical explanation, a geometry concept or a 3D scene brief. This Skill guides it to write a Python scene with ManimGX, inspect sampled frames and the timeline, correct unintended overlap or clipping, and export a video at your requested resolution and frame rate. The result includes the MP4 and its editable Python scene. ManimGX uses a Rust and wgpu renderer with a Python API based on Manim Community Edition.
Why we picked it
A task-focused Skill for turning an explanation into a renderable scene. It guides scene authoring, storyboard inspection and video delivery using the ManimGX engine. The preview is the upstream author’s helicoid-to-catenoid demonstration.
What it helps with
- Explain formulas, functions, geometry and changing values with animated diagrams.
- Create 3D objects, surfaces and camera movements for educational or technical videos.
- Iterate on an animation from its storyboard, then deliver an MP4 and reusable Python source.
From brief to finished video
Define the scene
Describe the idea, audience, duration, dimensions and frame rate. Use existing Python environment and install the rendering engine if needed.
Write editable Python
Author a Scene or ThreeDScene, arrange labels with their subjects, and animate changes through play and wait.
Inspect and refine
Generate the storyboard and timeline, inspect representative moments, and correct unintended overlaps or clipping. 3D inspection provides storyboards without 2D layout checks.
Render and deliver
Review the draft’s motion, render at the requested size and frame rate, and deliver both the video and Python source.
Rendering commands
After creating scene.py and installing the engine, inspect before rendering. The scene class in these commands is CircleArea, from the official Skill example.
| Command | What it helps with |
|---|---|
uv tool install --python 3.14 manimgx | Install the engine in an isolated tool environment. |
manimgx inspect scene.py CircleArea | Write a storyboard and report the timeline and sampled layout issues. |
manimgx render scene.py CircleArea -r 1280x720 --fps 30 -o draft.mp4 | Render a review video at 720p and 30 fps. |
Preview and practical limits
The preview is a compressed copy of the author’s full helicoid-to-catenoid film, linked below with its Python example. It illustrates the engine’s output; your agent creates a scene for your own brief.
Layout inspection samples selected moments; it does not guarantee every frame is free from overlap. Review the full video. Check ManimGX’s own reference for supported API behavior when adapting Manim CE code.
Official installable SkillManimGX API referenceAuthor’s original animated demonstrationEditable helicoid and catenoid scene
Before you use it
- An agent that can load SKILL.md, run shell commands, edit Python and inspect images. The skills installer requires Node.js and network access.
- Python 3.13 or newer and the manimgx package. Use an existing suitable Python environment, or install the engine with uv tool install --python 3.14 manimgx.
- A supported graphics backend: Metal on macOS, Direct3D 12 on Windows or Vulkan on Linux. Linux wheels also bundle a CPU rendering fallback.
- Text, math typesetting, fonts and video encoding are bundled. Optional generated narration needs a separately configured speech provider and may incur charges.
- The Skill and ManimGX’s own code are MIT-licensed. Bundled runtime components retain their own licenses. Agent or model usage follows your provider’s pricing.
Supported agents
Frequently asked questions
Does it produce a finished MP4?
Yes. With the ManimGX engine installed, the agent writes and renders a Python scene and delivers the video together with the scene source.
Do I need a separate LaTeX or FFmpeg installation?
The released ManimGX package includes text and math typesetting, fonts and video encoding. Follow the official installation instructions for your platform.
Is video generation free?
The Skill and local rendering engine are free to install. Agent/model usage, computing resources and any optional speech provider can have separate costs.
Sources & references
Original sources are linked below. Practical starting points are editorial suggestions from SkillHey.
