manim-math-animation

name	manim-math-animation
description	Use when the user wants animated math (handwritten-style equation writing, equation morphing between derivation steps, and emphasis) rendered with Manim, as a silent video clip that can stand alone or be spliced into a slides-to-video deck. The free, optional companion to slides-to-video for math lectures.
metadata	{"short-description":"Manim math-animation clips (write/morph equations) for lectures"}

Manim Math Animation

Render math animations with Manim from a simple JSON scene spec -- no hand-written Manim Python required. Output is a SILENT clip normalized to the slides-to-video canonical profile (resolution, fps, yuv420p, silent 48 kHz stereo audio), so it splices cleanly into a slides-to-video deck as an interlude, or stands alone.

This is the optional Manim companion to slides-to-video. Keep narration in the parent slides-to-video pipeline (one timing owner per segment): Manim owns only the animation timing and is rendered silent.

Windows Runtime Commands

$runtime = if ($env:AAS_RUNTIME_ROOT) { $env:AAS_RUNTIME_ROOT } elseif (Test-Path "$env:USERPROFILE\.codex\runtime") { "$env:USERPROFILE\.codex\runtime" } else { "$env:LOCALAPPDATA\ai-agents-skills\runtime" }
& "$runtime\run_skill.bat" "skills/manim-math-animation/run_manim_math_animation.bat" <args>
& "$runtime\run_skill.bat" "skills/manim-math-animation/run_manim_math_animation.ps1" <args>

When to use

• write an equation on screen in the 3Blue1Brown "being written" style (Write)
• morph one equation into the next derivation step (TransformMatchingTex)
• emphasize a step (indicate / circumscribe / flash / wiggle)
• produce a clip to drop into a slides-to-video lecture, or a standalone math clip

For plain narrated slides without authored animation, use slides-to-video directly. For genuine cursive handwriting, that is out of scope (no free Vietnamese-capable model); "handwriting" here means Manim Write of typeset math.

Setup (one time, explicit)

Manim needs a LaTeX distro (with dvisvgm + the standalone/preview packages

• cm-super), the cairo/pango dev libraries, and ffmpeg:

# Debian/Ubuntu
sudo apt-get install -y ffmpeg dvisvgm texlive texlive-latex-extra texlive-fonts-extra libcairo2-dev libpango1.0-dev build-essential

Then create the dedicated venv (installs Manim CE). Python 3.11+ receives Manim 0.20.1; Python 3.10 receives Manim 0.19.1, the latest compatible line for that interpreter.

bash /workspace/skills/manim-math-animation/run_manim_math_animation.sh setup
bash /workspace/skills/manim-math-animation/run_manim_math_animation.sh doctor

doctor reports readiness; it never installs anything.

Scene spec and subcommands

A scene spec is JSON, e.g.:

{
  "title": "Hàm số bậc hai",
  "equations": ["f(x) = x^2 + 2x + 1", "f(x) = (x + 1)^2"],
  "emphases": [{"at": 1, "type": "circumscribe"}],
  "run_time": 1.5, "width": 1920, "height": 1080, "fps": 30
}

• gen --spec spec.json --output scene.py -- generate the Manim script (no render).
• render --spec spec.json --output clip.mp4 [--quality=-qh] -- render + normalize to a silent, splice-ready clip. Pass the quality with = (e.g. --quality=-ql for a fast draft) because the value starts with a dash; omit it to use the default -qh.
• doctor / selftest -- environment check / offline smoke.

Rendering in OpenClaw (IMPORTANT)

Manim cannot run inside the OpenClaw sandbox (no build toolchain). Do NOT call render directly here — instead submit the render to the host job queue, which renders on the host and returns the clip:

exec: bash /workspace/skills/manim-math-animation/run_manim_job.sh --spec <spec.json> --output <clip.mp4> [--quality -ql|-qm|-qh] [--timeout 900]

Write the scene-spec JSON somewhere under /workspace (e.g. {{ PRIVATE_DATA_DIR }}/...) so the host worker can read it. The command prints the worker's JSON result (status: ok + clip) and leaves the .mp4 in place for sending. gen/doctor/ selftest still run locally in the sandbox.

Equations are LaTeX (math is language-neutral). Use the title (Pango Text) for Vietnamese/other-script prose; it renders with a Unicode font (Noto by default).

Integration with slides-to-video

Render a clip here, then treat it as one slide's visual in slides-to-video: the parent measures the clip duration with ffprobe and concatenates it like any other slide, synthesizing that segment's narration via its own TTS ladder. Because the Manim clip is silent and already on the canonical profile, the concat stays drift-free. See references/integration.md.