peter-gy/manim-marimo

manim-marimo brings Manim animations to marimo notebooks through marimo-studio. The notebook computes the values. A Manim scene in a Studio view animates them, and Studio renders the movie again whenever a value changes.

import numpy as np
from manimlib import Scene, ShowCreation, VMobject
from manim_marimo import value


class Animation(Scene):
    def construct(self):
        points = np.asarray(value("trajectory", default=[[0, 0], [1, 1]]), dtype=float)
        curve = VMobject()
        curve.set_points_as_corners(np.column_stack((points, np.zeros(len(points)))))
        self.play(ShowCreation(curve), run_time=3)

trajectory is a variable in the notebook. The view's page plays the movie beside the notebook's live controls, and output() and cell() place notebook figures inside the scene.

Get started

manim-marimo renders scenes in the notebook's pixi environment, which installs the OpenGL and Pango libraries ManimGL renders with from conda-forge. Declare it in the notebook's script metadata:

# /// script
# requires-python = ">=3.10,<3.15"
# dependencies = [
#     "manim-marimo[render] @ git+https://github.com/peter-gy/manim-marimo",
# ]
#
# [tool.pixi.workspace]
# channels = ["conda-forge"]
#
# [tool.pixi.dependencies]
# manimpango = ">=0.6"
# ///

uvx --with "manim-marimo @ git+https://github.com/peter-gy/manim-marimo" \
  marimo edit analysis.py --sandbox=pixi --watch

In Studio, click Add view and choose Manim animation. Getting started shows the complete metadata, including the Mesa tables that render on Linux servers without a display.

Author with a coding agent

manim-marimo ships an agent skill beside marimo-studio's. Paste this into Claude Code, Codex, or another terminal agent:

Read the marimo-studio and manim-marimo skills with
`uvx --with "manim-marimo @ git+https://github.com/peter-gy/manim-marimo" agent-plugins read <name>`,
then create a notebook that explains Fourier series and an Animation view
that draws a square wave from rotating vectors.

Examples

• A notebook is a dataflow graph: the notebook parses a small notebook, sorts its cells topologically, and finds what each slider reruns. The scene draws those results.
• Fourier drawing: rotating vectors draw manim-marimo's mark and TeX's italic m from Fourier coefficients the notebook computes.
• Convolution of dice: a joint probability table folding into the distribution of a sum, ending on the notebook's own chart.
• Simulating an epidemic: two hundred people simulated in the notebook, after 3Blue1Brown's Simulating an epidemic. Pause the movie and point at anyone to trace them to the notebook row that simulated them.

pixi run edit dataflow

Documentation

Guide · Examples · Reference · Contributing

License

Apache License 2.0. The epidemic example's scene.py adapts code from 3b1b/videos and shares its CC BY-NC-SA 4.0 license.