This repository provides AI agent skills for Manim, a Python library for creating mathematical animations, enabling AI assistants to generate 3Blue1Brown-style visuals. It targets developers and researchers looking to integrate programmatic animation capabilities into AI workflows, offering battle-tested best practices and code examples for both Manim Community Edition and ManimGL.

How It Works

The project offers two distinct sets of AI agent skills, one for Manim Community Edition (CE) and another for ManimGL (Grant Sanderson's original version). These skills are designed to be added to AI coding agents via npx skills add, automatically activating when Manim-related code is detected. The approach leverages the Agent Skills open standard to provide domain-specific knowledge, code examples, and best practices tailored to each Manim framework, acknowledging their fundamental incompatibility.

Quick Start & Requirements

• Install Command: npx skills add adithya-s-k/manim_skill/skills/manimce-best-practices and/or npx skills add adithya-s-k/manim_skill/skills/manimgl-best-practices.
• Prerequisites: Python 3.7+, FFmpeg (for video encoding), and LaTeX (TeX Live, MiKTeX, or MacTeX) are required for both Manim versions.
• Dependencies: uv is recommended for package management.
• Links:
  • Manim Community Edition: https://github.com/ManimCommunity/manim, Docs: https://docs.manim.community/
  • ManimGL: https://github.com/3b1b/manim

Highlighted Details

• AI Agent Integration: Seamlessly enhances AI coding assistants with Manim expertise.
• Comprehensive Examples: Provides working, tested code examples and best practices for specific Manim features.
• Version Differentiation: Clearly outlines key differences and usage patterns between Manim CE and ManimGL.
• Testing Framework: Includes tests for all skill examples, runnable via uv run python tests/<version>/test_all_skills.py.

Maintenance & Community

The repository acknowledges the foundational work of Grant Sanderson (3Blue1Brown) and the Manim Community contributors. Community resources include the Manim Community Discord (https://www.manim.community/discord/) and GitHub repositories for both Manim versions.

Licensing & Compatibility

The educational materials and code examples within this repository are licensed under the MIT License. However, users must be aware that the underlying Manim Community Edition and ManimGL frameworks are governed by their own respective licenses, which may have different terms and restrictions.

Limitations & Caveats

Manim Community Edition and ManimGL are distinct and incompatible; code written for one will not work with the other without modification. Installation requires specific external dependencies (FFmpeg, LaTeX) that can be complex to set up. The README includes a troubleshooting section for common installation and usage issues. Parallel testing with many workers can lead to out-of-memory errors.