English | 简体中文
∫ ∑ ∂ ∞
🎬 AI-Powered Mathematical Animation Generator
Making mathematical animation creation simple and elegant, powered by Manim and large language models
Overview • Examples • Quick Start • Technology • Deployment • Additions • License • Maintenance
I am happy to introduce my new project, ManimCat. It is, after all, a cat.
Built on top of manim-video-generator, ManimCat is now a much broader AI-assisted creation system for math teaching visuals rather than just a single generation flow.
It is designed for classroom explanation, worked-example breakdowns, and visual reasoning tasks. You can use natural language to generate, modify, rerender, and organize Manim-based teaching content with both video and image outputs.
The project currently has two distinct modes: Workflow Mode for direct generation and rendering, and Agent Mode for Studio-based collaborative work with longer-lived sessions, review, and iteration.
Prove that $1/4 + 1/16 + 1/64 + \dots = 1/3$ using a beautiful geometric method, elegant zooming, smooth camera movement, a slow pace, at least two minutes of duration, clear logic, a creamy yellow background, and a macaroon-inspired palette.
default.mp4
▲ Generated with BGM · Geometric Series Proof · ManimCat
npm install
cd frontend && npm install
cd ..
npm run devOpen http://localhost:3000. For environment variables and deployment-specific setup, see the deployment guide.
- Product modes: Workflow mode for direct generation, Agent mode for Studio-based collaborative work
- Backend: Express + TypeScript, Bull + Redis, OpenAI-compatible upstream routing, Studio agent runtime, optional Supabase history storage
- Frontend: React 19, Vite, Tailwind CSS, classic generator UI, Studio workspace shell
- Realtime layer: polling for Workflow jobs, Server-Sent Events for Agent sessions, permission requests, and task updates
- Rendering runtime: Python + Manim Community Edition, LaTeX,
ffmpeg - Deployment: Docker / Docker Compose, Hugging Face Spaces
flowchart LR
classDef ui fill:#F6F7FB,stroke:#455A64,color:#263238,stroke-width:1.2px;
classDef logic fill:#FFF8E1,stroke:#A1887F,color:#4E342E,stroke-width:1.2px;
classDef api fill:#E8F5E9,stroke:#5D8A66,color:#1B4332,stroke-width:1.2px;
classDef state fill:#E3F2FD,stroke:#5C6BC0,color:#1A237E,stroke-width:1.2px;
classDef output fill:#FCE4EC,stroke:#AD5C7D,color:#6A1B4D,stroke-width:1.2px;
U[User prompt] --> P1
P1[Classic UI] --> P2[Problem framing]
P1 --> P3[Generate / Modify requests]
P2 --> A1[Workflow APIs]
P3 --> A1
A1 --> R1[Upstream routing + AI generation]
R1 --> C1[Static checks + retry / patch loop]
C1 --> B1[Queue + job state]
B1 --> B2[Render pipeline]
B2 --> O1[Video / images / code / timings]
P1 -. polling / cancel .-> B1
O1 --> P1
class P1 ui;
class P2,P3,R1,C1 logic;
class A1 api;
class B1,B2 state;
class O1 output;
flowchart LR
classDef ui fill:#F6F7FB,stroke:#455A64,color:#263238,stroke-width:1.2px;
classDef runtime fill:#FFF8E1,stroke:#A1887F,color:#4E342E,stroke-width:1.2px;
classDef api fill:#E8F5E9,stroke:#5D8A66,color:#1B4332,stroke-width:1.2px;
classDef state fill:#E3F2FD,stroke:#5C6BC0,color:#1A237E,stroke-width:1.2px;
classDef event fill:#FCE4EC,stroke:#AD5C7D,color:#6A1B4D,stroke-width:1.2px;
U[User instruction] --> S1
S1[Studio UI] --> A1[Session / run APIs]
A1 --> R1[Studio runtime service]
R1 --> K1[Manim Studio / Plot Studio]
K1 --> G1[Builder, Designer, Reviewer]
G1 --> T1[Tools, skills, render / review actions]
R1 --> S2[Session / run / task / work state]
R1 --> E1[SSE events + permissions]
S2 --> S1
E1 --> S1
class S1 ui;
class A1 api;
class R1,K1 runtime;
class G1,T1,S2 state;
class E1 event;
For environment variables, deployment modes, and upstream-routing examples, see the deployment guide.
Please see the deployment guide.
This project is a substantial rework built on top of the original foundation. The main additions I personally designed and implemented are:
- Added a dedicated image workflow alongside video generation
- Added
YON_IMAGEanchor-based segmented rendering for multi-image outputs - Added two-stage AI generation: a concept designer produces a scene design, then a code generator writes the Manim code
- Added a static analysis guard (
py_compile+mypy) that checks generated code before rendering, with AI-powered auto-patching for up to 3 passes - Added AI-driven code retry: when a render fails, the error is fed back to the model to regenerate and re-render automatically
- Added rerender-from-code and AI-assisted modify-and-render flows
- Added stage timing breakdown shared by both image and video jobs
- Added background music mixing for rendered videos
- Added render-failure event collection and export for debugging and reliability work
- Rebuilt the frontend as a separate React + TypeScript + Vite application
- Added problem framing before generation to help structure user requests
- Added reference image upload support
- Added a unified workspace for generation history and usage views
- Added a usage metrics dashboard with daily charts, success rates, and timing breakdowns
- Added a prompt template manager for viewing and overriding system prompts per role
- Added dark / light theme toggle
- Added a waiting-state 2048 mini-game
- Added a refreshed visual style, settings panels, and provider configuration flows
- Reworked the backend around Express + Bull + Redis
- Added retry, timeout, cancellation, and status-query flows
- Added support for third-party OpenAI-compatible APIs and custom provider configuration
- Added server-side upstream routing by ManimCat key
- Kept optional multi-profile frontend provider rotation for local use
- Added optional Supabase-backed persistent history storage
- Added a separate Studio mode driven by an agent runtime rather than the classic generation flow
- Added session, run, task, work, and work-result state models for long-lived agent interactions
- Added builder, designer, and reviewer roles inside the Studio agent system
- Added workspace tools, render tools, local skills, and subagent orchestration
- Added Server-Sent Events for live Studio updates plus permission request / reply handling
- Added Studio review, pipeline, work, and permission panels in the frontend
Licensing details are defined in LICENSE_POLICY.md (Chinese) and LICENSE_POLICY.en.md (English).
- Third-party attribution and notices:
THIRD_PARTY_NOTICES.md - Chinese third-party notices:
THIRD_PARTY_NOTICES.zh-CN.md - Contribution agreement:
CLA.md - Contribution guide:
CONTRIBUTING.md
This project uses a CLA so the maintainer can keep commercial-use authorization under a single workflow, instead of requiring separate consent from every contributor each time.
This is not a statement of "project commercialization" as a company path. The maintainer remains a developer, the project stays open source, and the project stance is anti-monopoly.
Any commercial authorization income is intended to be reinvested into project development itself, including support for major contributors and community activities. For small companies that are open-source-friendly, authorization terms and fees are expected to be symbolic.
Because my time is limited and I am an independent hobbyist rather than a full-time professional maintainer, I currently cannot provide fast review cycles or long-term maintenance for external contributions. Pull requests are welcome, but review may take time.
If you have good suggestions or discover a bug, feel free to open an Issue for discussion. I will improve the project at my own pace. If you want to make large-scale changes on top of this work, you are also welcome to fork it and build your own version.
If this project gave you useful ideas or helped you in some way, that is already an honor for me.
If you like this project, you can also buy the author a Coke 🥤
Mainland China:
International:
Thank you. Your support gives me more energy to keep maintaining the project.