manim-skill is a reusable skill package for creating technical animations with ManimCE. It works as a Codex native skill and as an agent-agnostic playbook for any AI coding assistant that reads markdown.
You describe the animation you want. The agent handles the rest — pattern selection, code generation, rendering, troubleshooting, and iteration.
When provided with SKILL.md and relevant references/ docs, AI agents can autonomously:
| Step | What the agent does |
|---|---|
| 1. Pattern selection | Choose the right animation type (bars, graphs, diagrams, 3D, camera moves) |
| 2. Code generation | Write correct ManimCE scene code from your description |
| 3. Rendering | Execute render commands with appropriate quality flags |
| 4. Troubleshooting | Fix FFmpeg, TeX, font, and environment issues automatically |
| 5. Iteration | Refine timing, colors, and layout until the output matches your intent |
| Agent / IDE | How to use |
|---|---|
| Codex | Install as a native skill folder (~/.codex/skills/...) |
| Claude Code / Cursor / Cline | Use SKILL.md + references/ as project instructions or prompt context |
| Other agents | Use this repository as a structured Manim playbook |
Option 1: Codex native skill
# macOS / Linux
python ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
--repo awesome-skills/manim-skill \
--path . \
--name manim-skill# Windows PowerShell
python $HOME\.codex\skills\.system\skill-installer\scripts\install-skill-from-github.py `
--repo awesome-skills/manim-skill `
--path . `
--name manim-skillOption 2: Standalone playbook
git clone https://github.com/awesome-skills/manim-skill.git
cd manim-skillThen feed SKILL.md and relevant references/ files into your agent workflow.
# Fast preview (480p)
uv run --with manim manim -pql examples/quicksort.py QuickSortBars
# GIF export (720p)
uv run --with manim manim -qm --format=gif examples/quicksort.py QuickSortBars
# 4K showcase (2160p60)
manim -qk examples/quicksort.py QuickSortBars
| File | Lines | Description |
|---|---|---|
| SKILL.md | ~380 | Core instructions — loaded on skill activation |
| references/mobjects.md | ~860 | Mobject types, positioning, styling, grouping |
| references/animations.md | ~620 | Animation classes, timing, chaining, transforms |
| references/advanced.md | ~890 | 3D scenes, camera, ValueTracker, updaters, custom anims |
| references/blog-patterns.md | ~410 | Blog-ready patterns: hero banners, step-by-step, comparisons |
Total: 3,100+ lines of ManimCE guidance, loaded progressively per topic.
| File | What it demonstrates |
|---|---|
examples/basic_scene.py |
Minimal scene setup, shapes, text, basic animations |
examples/quicksort.py |
Algorithm visualization with bar charts and step labels |
examples/flowchart.py |
Flowchart construction with arrows and styled boxes |
examples/state_diagram.py |
State machine with transitions and highlighting |
manim-skill/
|
+-- SKILL.md # Core skill instructions (~380 lines)
+-- README.md # English documentation
+-- README.zh-CN.md # Chinese documentation
|
+-- examples/ # Ready-to-run animation scenes
| +-- basic_scene.py # Minimal hello-world scene
| +-- quicksort.py # Algorithm bar chart animation
| +-- flowchart.py # Flowchart with arrows and boxes
| +-- state_diagram.py # State machine visualization
|
+-- references/ # Progressive reference guides
| +-- mobjects.md # Shapes, text, tables, graphs, grouping
| +-- animations.md # Create, Transform, Fade, timing control
| +-- advanced.md # 3D, camera, ValueTracker, updaters
| +-- blog-patterns.md # Blog-ready animation recipes
|
+-- assets/ # Media assets
+-- banner.svg # Repository banner
+-- quicksort-preview.mp4 # Preview video
+-- quicksort-preview.png # Static fallback image
- ManimCE v0.19+ — No external FFmpeg needed (uses pyav internally)
- Progressive loading — Only load the reference docs relevant to your task
- Windows-friendly — PowerShell setup,
uv runone-liners, font fallback tips - TeX-free fallback — Numeric labels and
Text()alternatives when LaTeX is unavailable - Production-ready — Quality presets from 480p preview to 4K60 showcase
Contributions are welcome! Priority areas:
- New algorithm animation templates
- Better cross-platform troubleshooting
- More agent integration examples
- Framework-specific patterns (D3-style, Matplotlib-style)
- Translations
MIT © awesome-skills