Wing900/ManimCat

What it solves

ManimCat is an AI-assisted creation system designed to generate mathematical visuals for teaching and classroom explanations. It removes the technical barrier of writing complex animation code by allowing users to create both animated videos and static figures using natural language prompts.

How it works

The system operates using two primary modes and two rendering engines:

• Workflow Mode: A direct path for fast generation where a prompt is processed through a two-stage AI pipeline (concept design then code generation), checked by a static analysis guard, and rendered into a final output.
• Agent Mode: A collaborative studio environment with longer-lived sessions. It uses a multi-role agent system (Builder, Designer, Reviewer) to iteratively refine visuals through tasks and reviews.
• Engines: It uses Manim for timeline-based animations and matplotlib for static charts and teaching figures.

Who it’s for

It is primarily built for educators, teachers, and anyone creating mathematical content who needs high-quality visual reasoning tasks, worked-example breakdowns, and classroom explanations without manually coding the visuals.

Highlights

• Dual-Engine Support: Seamlessly switches between Manim for animation and matplotlib for static images.
• AI-Powered Auto-Patching: Includes a static analysis guard that detects code errors and uses AI to automatically patch them before rendering.
• Two-Stage Generation: Separates scene design from code implementation to improve visual quality.
• Collaborative Studio: An agent-driven workspace that supports iterative editing, session state, and task-based refinement.
• Integrated Tooling: Features background music mixing for videos and reference image upload support.