🎬 Manim Animation Studio

Professional Desktop Environment for Creating Mathematical Animations

A powerful, feature-rich desktop application for creating stunning mathematical animations using Manim Community Edition. Built with Python, PyWebView, and modern web technologies.

✨ Features

🎨 Professional Code Editor

Monaco Editor (VS Code's editor) with full Python syntax highlighting
basedpyright IntelliSense — real-time type checking, completions, hover docs, and diagnostics
Go-to-Definition (F12), Type Definition, and Find All References (Shift+F12) via LSP
VS Code-quality signature help — live parameter hints for Python builtins and all Manim classes
Static completions for all Manim classes, animations, mobjects, colors, and constants (instant, no LSP required)
Scene Outline Panel — tree view of all classes and methods with click-to-navigate
Editor Bookmarks (Ctrl+Shift+K) — bookmark lines for quick navigation
Zen Mode (F11) — distraction-free fullscreen editing
Drag-and-drop .py files onto the editor to open them
Line numbers, code folding, and multiple cursors
Customizable font size and themes

🤖 AI Edit Panel (Codex + Claude Code)

Send your code with a natural language prompt for intelligent edits
Claude Code (Beta) — structured stream-json integration, no API key needed
OpenAI Codex CLI — full JSONL streaming with codex exec --json
AI Agent Mode — autonomous generate → render → screenshot → review → fix loop
- Works with both Claude and Codex providers
- Visual QA review examines every frame for issues
- Auto-iterates until the animation looks correct
Continuous Chat — multi-turn conversations with session memory (--session-id / --resume)
Token-Optimized — AI reads files from disk instead of prompt embedding, session caching saves 70-80% on subsequent turns, screenshots downscaled to reduce image tokens
Image upload — attach screenshots or reference images for the AI to understand your intent
Web Search toggle — let the AI reference live web results while editing
Model selector — pick from Claude Opus/Sonnet/Haiku, GPT-5.3 Codex, and more
Side-by-side diff review — Accept or Reject changes with one click
Inline autocomplete — ghost-text code completions powered by Claude Haiku
Fix code errors directly from the diagnostics panel
Premium glass-morphism UI with gradient buttons and slide-in panel

🎙️ Auto Narration (Kokoro TTS)

Add narrate("Your text here") anywhere in your Manim code
TTS audio is automatically generated and merged with the video on render or preview
Auto-subtitles (CC) — WebVTT captions appear in the preview player
Powered by Kokoro ONNX (82M params, ~310MB model, auto-downloaded on first use)
Multiple voices — Heart, Bella, Nicole, Sarah, Adam, Michael, Emma, George
Optional narrate[kokoro] package available in the setup wizard

⚡ Dual Render Modes

Quick Preview (F6): Fast, low-quality preview (480p, 15fps) for rapid iteration
Final Render (F5): High-quality output up to 8K resolution at 120fps
GPU acceleration support (OpenGL renderer)
Real-time progress tracking with live terminal output
Render History — persistent log of all renders with replay, open, and delete

📁 Integrated Asset Management

Drag & drop file uploads
Support for videos, images, fonts, audio, and subtitles
Built-in asset preview with media player
Organized asset library with file metadata
Easy access to custom fonts and media files

📦 Visual Package Manager

Install Python packages with one click
Check for package updates
Uninstall packages safely
View installed packages and versions
Auto-detection of missing required packages on startup
No terminal commands needed

🖥️ Full-Featured Terminal

Real Windows cmd.exe integration with xterm.js
Auto-activated virtual environment
Copy/paste support (Ctrl+Shift+C/V)
Persistent session across renders
Color-coded output for better readability

🖥️ CLI & MCP Server (Codex Plugin)

Same EXE, dual mode — double-click for GUI, run from terminal for CLI
ManimStudio render scene.py --quality 1080p --width 1920 --height 1080 --fps 60 — headless render
ManimStudio validate scene.py — syntax check and scene class detection
ManimStudio presets — list all quality presets with resolutions
ManimStudio mcp — start MCP server (stdio) for OpenAI Codex integration
Full resolution control — quality presets (120p–8K) plus custom --width / --height override
MCP tools exposed: render_manim_animation, check_render_status, validate_scene, list_quality_presets
Uses the same manim_studio_default venv as the GUI — no separate setup
Register in Codex: add [mcp_servers.ManimStudio] to ~/.codex/config.toml

🎯 Smart Workflow Features

Auto-save backups: Never lose your work
Unsaved changes warning: Prompted before opening a new file
File recovery: Restore from auto-saved versions
Auto-open output folder: Jump directly to saved renders
Live preview: See your animation immediately after render
Screenshot save: Save preview frames with a native Save As dialog
Settings persistence: Your preferences are remembered
Command Palette (Ctrl+Shift+P) — quick launcher for all actions

🌓 Modern UI/UX

Dark/Light theme toggle
Manim Color Picker — visual palette of all Manim color constants with one-click insert
Keyboard Shortcuts Modal (Ctrl+/) — searchable reference of all shortcuts
Responsive design with DPI awareness
Glassmorphic design elements
Smooth animations and transitions
Collapsible render controls sidebar
Toast notifications for all actions

📸 Screenshots

Main Workspace

The main interface features a split view with Monaco code editor on the left and live preview on the right, with an integrated terminal at the bottom.

Asset Manager

Browse, upload, and preview all your media files in one organized location with drag-and-drop support.

Package Manager

Install and manage Python packages through an intuitive visual interface without touching the command line.

🚀 Getting Started

Prerequisites

Windows 10/11 (64-bit)
Python 3.8+ installed and in PATH
4GB RAM minimum (8GB recommended)
1GB free disk space

Installation

Download the latest release
```
Download from GitHub Releases
```
Extract the archive
```
Extract to your preferred location
```

Run the application

Double-click app.exe (or python app.py from source)

First-time setup (automatic)
- Virtual environment creation
- Manim, manim-fonts, and basedpyright installation
- LaTeX detection and setup
- Assets folder initialization

Building from Source

# Clone the repository
git clone https://github.com/yu314-coder/manim_app.git
cd manim_app

# Install host dependencies
pip install pywebview

# Run the application
python app.py

Building Executable with Nuitka

# Activate build environment
python -m venv venv_nuitka
venv_nuitka\Scripts\activate

# Install Nuitka
pip install nuitka

# Build executable
python build_nuitka.py

# Output will be in dist_nuitka/

CLI Usage

The same executable (or python app.py) works as both GUI and CLI:

# Headless render with quality preset
ManimStudio.exe render scene.py --quality 1080p --fps 60

# Render with custom resolution (overrides preset width/height)
ManimStudio.exe render scene.py --quality 720p --width 800 --height 600 --fps 24

# Render to a specific output directory
ManimStudio.exe render scene.py -q 4K -o C:\renders\my_video

# Validate code without rendering
ManimStudio.exe validate scene.py

# List all quality presets
ManimStudio.exe presets

# Start MCP server for Codex integration
ManimStudio.exe mcp

From source, replace ManimStudio.exe with python app.py:

python app.py render scene.py --quality 720p --width 1280 --height 720
python app.py mcp

Codex MCP Integration

Register ManimStudio as an MCP server in ~/.codex/config.toml:

# If using the compiled EXE:
[mcp_servers.ManimStudio]
command = "C:\\path\\to\\ManimStudio.exe"
args = ["mcp"]

# If running from source:
[mcp_servers.ManimStudio]
command = "python"
args = ["C:\\path\\to\\app.py", "mcp"]

Once registered, Codex can call these MCP tools:

Tool	Description
`render_manim_animation`	Render Manim code with quality/width/height/fps/format controls
`check_render_status`	Check if a render output file exists
`validate_scene`	Syntax check + scene class name extraction
`list_quality_presets`	List all presets with their default resolutions

📖 Quick Start Guide

Creating Your First Animation

Launch Manim Studio
- The app opens with a default scene template

Write Your Code

from manim import *

class MyScene(Scene):
    def construct(self):
        circle = Circle()
        self.play(Create(circle))
        self.wait()

Preview (F6)
- Click "Preview" or press F6
- Get a fast low-quality preview in seconds
Final Render (F5)
- Click "Render" or press F5
- Choose quality settings in the sidebar
- Save your high-quality video when complete
- Output folder opens automatically after saving

Adding Narration

from manim import *

class NarratedScene(Scene):
    def construct(self):
        narrate("Let me show you a circle.")
        circle = Circle(color=BLUE)
        self.play(Create(circle))
        narrate("Now let's transform it into a square.")
        self.play(Transform(circle, Square(color=RED)))
        self.wait()

Press F6 or F5 — TTS audio is generated and merged with the video automatically. Subtitles appear in the preview player.

Keyboard Shortcuts

Action	Shortcut
Render Animation	`F5`
Quick Preview	`F6`
Stop Render	`Esc`
Save File	`Ctrl + S`
New File	`Ctrl + N`
Open File	`Ctrl + O`
Command Palette	`Ctrl + Shift + P`
Keyboard Shortcuts	`Ctrl + /`
Zen Mode	`F11`
Toggle Bookmark	`Ctrl + Shift + K`
AI Edit Panel	`Ctrl + Shift + E`
Go to Definition	`F12`
Find All References	`Shift + F12`
Screenshot Preview	`Ctrl + Shift + S`

🎨 Using Assets

Adding Custom Fonts

Upload Font
- Go to Assets tab
- Drag & drop your .ttf or .otf file

Use in Code

from manim import *

class CustomFontExample(Scene):
    def construct(self):
        register_font("YourFont.ttf")
        text = Text("Hello!", font="YourFont")
        self.play(Write(text))

Adding Images/Videos

Upload Media
- Drag files to Assets tab dropzone
- Supported: MP4, PNG, JPG, SVG, GIF

Use in Animation

from manim import *

class ImageExample(Scene):
    def construct(self):
        img = ImageMobject("myimage.png")
        self.play(FadeIn(img))

Adding Audio

from manim import *

class AudioExample(Scene):
    def construct(self):
        self.add_sound("music.mp3")
        circle = Circle()
        self.play(Create(circle))

⚙️ Configuration

Render Settings

Quality Presets (GUI, CLI --quality, and MCP quality param):

Preset	Resolution	Default FPS	CLI Flag
120p	214×120	15	`-q 120p`
240p	426×240	15	`-q 240p`
480p	854×480	15	`-q 480p`
720p	1280×720	30	`-q 720p`
1080p	1920×1080	60	`-q 1080p`
1440p	2560×1440	60	`-q 1440p`
4K	3840×2160	60	`-q 4K`
8K	7680×4320	60	`-q 8K`

Use --width / --height (CLI) or width / height (MCP) to override the preset resolution.

Output Formats:

MP4 (H.264 video)
MOV (QuickTime)
GIF (Animated)
PNG (Image sequence)

GPU Acceleration

Toggle GPU rendering in the header toolbar:

GPU: OFF = Cairo renderer (CPU, compatible)
GPU: ON = OpenGL renderer (GPU, faster but may have quirks)

Editor Settings

Access via Settings (⚙️ icon):

Font size (12-20px)
Default save location
Auto-save after render
Auto-open output folder
Manim cache settings

📂 File Structure

manim_app/                    # Source / EXE directory
├── app.py                    # Main app (GUI + CLI dispatcher)
├── cli.py                    # CLI & MCP server (headless render)
├── ai_edit.py                # AI Edit module (Claude + Codex)
├── narration_addon.py        # Kokoro TTS narration
├── build_nuitka.py           # Nuitka build script
├── prompts/                  # AI prompt templates (.md)
└── web/                      # Frontend (HTML/CSS/JS)

C:\Users\<you>\.manim_studio\   # User data directory
├── assets\              # Your uploaded files (fonts, images, audio)
├── lsp_workspace\       # basedpyright workspace (auto-managed)
├── media\               # Manim cache and temp files
├── render\              # High-quality render output (temporary)
├── preview\             # Quick preview output (temporary)
├── autosave\            # Auto-saved code backups
├── venvs\
│   └── manim_studio_default\  # Python virtual environment (shared by GUI & CLI)
└── settings.json        # Your app preferences

Note: Render outputs are temporary. Use the "Save" button after rendering to save them permanently to your chosen location. The output folder will open automatically in Windows Explorer after saving.

🐛 Troubleshooting

Rendering Fails

Problem: Render shows errors in terminal

Solutions:

Check syntax errors (red underlines in editor)
Ensure all imported files exist in assets folder
Try Preview (F6) first to catch errors quickly
Check LaTeX status in System tab
Review terminal output for specific error messages

Custom Fonts Not Working

Problem: Font doesn't appear in animation

Solutions:

Ensure font file is uploaded to Assets folder
Use register_font("YourFont.ttf") before using it
Reference font by name without extension: font="YourFont"
Check that font file is valid TrueType (.ttf) or OpenType (.otf)

LaTeX Not Working

Problem: LaTeX formulas fail to render

Solutions:

Check LaTeX status indicator (header, right side)
Install MiKTeX or TeX Live if not present
Restart application after LaTeX installation
Use raw strings for LaTeX: r"$\frac{1}{2}$"

IntelliSense Not Working

Problem: No type checking or smart completions from basedpyright

Solutions:

Wait ~2 seconds after the editor loads — IntelliSense starts in the background
Check that basedpyright is installed (Packages tab)
If missing, the app will prompt you to install it on startup
A "IntelliSense ready" toast appears when LSP is active

Package Installation Fails

Problem: Cannot install Python packages

Solutions:

Check terminal output for specific error
Ensure package name is spelled correctly
Try installing via terminal manually
Check internet connection
Some packages may require system dependencies

Video Won't Play in Preview

Problem: Preview panel shows no video after render

Solutions:

Check that render completed successfully (terminal logs)
Try refreshing Assets tab
Ensure video file exists and is not corrupted

🤝 Contributing

Contributions are welcome! Here's how you can help:

Report Bugs: Open an issue with detailed reproduction steps
Suggest Features: Describe your use case and proposed solution
Submit Pull Requests: Fork, branch, code, test, and PR
Improve Documentation: Fix typos, add examples, clarify instructions

Development Setup

# Clone and setup development environment
git clone https://github.com/yu314-coder/manim_app.git
cd manim_app

# Install host dependencies
pip install pywebview

# Run in development mode
python app.py

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

Manim Community - The amazing animation engine
Monaco Editor - VS Code's powerful editor
basedpyright - Python language server for IntelliSense
OpenAI Codex CLI - OpenAI's AI coding agent
Kokoro ONNX - Fast TTS with Kokoro and ONNX Runtime
PyWebView - Native Python desktop apps
xterm.js - Terminal emulator for the web
Font Awesome - Beautiful icons

📞 Support

Documentation: Check the built-in Help modal (? icon)
Issues: GitHub Issues
Discussions: GitHub Discussions
Manim Help: Manim Community Discord

🗺️ Roadmap

Upcoming Features

CLI & MCP Server: Headless render + Codex plugin via MCP tools
Live Preview Mode: See animations update in real-time as you type
Scene Browser: Visual navigation between multiple scenes
Animation Templates: Pre-built scenes for common use cases
Export Presets: Save and reuse render configurations
macOS/Linux Support: Cross-platform compatibility
Timeline Editor: Visual animation timeline

📊 System Requirements

Minimum

OS: Windows 10 (64-bit)
CPU: Dual-core 2.0 GHz
RAM: 4GB
Storage: 1GB free space
Display: 1280×720

🎓 Learning Resources

Manim Tutorials

Python Resources

⭐ Star History

If you find this project useful, please consider giving it a star on GitHub!

📝 Changelog

v1.1.3.0 (Latest)

New features:

✨ F-09 Agent Memory — per-project style.md auto-injected as a system preamble on every AI edit/agent turn. Editable via the 🧠 brain icon in the AI Edit panel header. Heuristic detects user corrections (e.g. "no, use BLUE instead of RED") and proposes memory updates with one-click accept/dismiss. Prefix #ignore-style on a single message to skip the preamble for that turn.
✨ F-03 Visual Diff — pHashes every rendered frame, diffs against the baseline render, opens an A/B review modal when drift > 1%. Accept (promotes new baseline), Revert, Block per render. Keyboard shortcuts J/K next/prev flagged frame, A accept, R revert, B block.
✨ Multi-scene combined render — files with 2+ Scene classes pop a picker. Tick checkboxes + "Render selected" runs ONE manim subprocess with all scene names as positional args, then ffmpeg -f concat -c copy stitches the outputs into one standalone combined_*.mp4.
✨ Scene picker modal — auto-shows when rendering/previewing a file with multiple Scene subclasses. Single-click a row to render just that scene; checkboxes + "Render selected" for batch; "Select all" button for one-click full-file render.
✨ CJK auto-template — files with Chinese / Japanese / Korean characters inside MathTex(...) / Tex(...) get the xelatex + ctex template auto-injected. Works around Manim 0.20.1's tempconfig({}) clearing _tex_template between scenes by subclassing MathTex/Tex to bake tex_template=ctex into every constructor.
✨ Auto-discovery of AI models — Claude models auto-detected by scanning ~/.claude/projects/*/*.jsonl session history; Codex models pulled from the authoritative ~/.codex/models_cache.json cache (with full metadata: slug, display name, reasoning levels, speed tiers, visibility). New models that OpenAI / Anthropic ship surface in the dropdown automatically — no Manim Studio update required. Discovered models appear under a "Discovered" tier with an orange 🔍 magnifier icon to distinguish from hardcoded defaults.
✨ Claude 4.7 family — Opus 4.7, Sonnet 4.7, Haiku 4.7 are now the default tier for edit / agent / autocomplete / review / complex tasks. 4.6 and 4.5 remain selectable under a "(previous)" suffix. Fallback chain: Opus 4.7 → Sonnet 4.7 → Haiku 4.7 → Haiku 4.5.
✨ AST-based scene detection — fixed the long-standing "only first scene detected" bug. extract_scene_name now uses Python's ast module to find every Scene subclass, handling tricky base-class shapes (typing.Generic[T], metaclass parents, mixed bases). Backward-compatible regex fallback for files in the middle of an edit.
✨ list_scenes MCP tool — Codex can list every Scene-subclass class in a Manim file (name, line, parent) before invoking render_manim_animation.

Bug fixes:

🐛 Multi-scene render no longer wipes preview folder between iterations. Was a per-scene loop that called clear_preview_folder() between scenes; the previous scene's temp file got deleted before manim could read it. Now a single subprocess + ffmpeg concat.
🐛 Agent waits for in-flight preview before firing its own. Was hanging forever when quickPreview() got silently rejected by if (job.running) return. Now polls job.running to clear (up to 5min) before triggering, plus a 10-min safety timeout that synthesises render_error to break out of _ai_agent_wait cleanly.
🐛 Preview-folder race condition fixed — double-clicking Preview rapidly wiped the in-flight previous render's temp .py (FileNotFoundError: temp_preview_*.py not found). clear_preview_folder() now skips files modified in the last 60 seconds.
🐛 LaTeX errors no longer truncated — error capture used to grab only 3 lines / 200 chars, hiding the actual root cause. Now scans up to 40 traceback lines, stops at the closing ErrorType: line, returns up to 2000 chars to the UI, and dumps the full content to backend stdout.
🐛 Memory modal no longer overlapped by preview video — Chromium's GPU video compositor painted over HTML regardless of z-index. Modal now sets visibility:hidden on the <video> element while open (full GPU-layer removal), plus transform: translateZ(0); isolation: isolate on the modal itself as defense-in-depth.

Removed from toolbar UI (backends remain on disk for future re-introduction):

🗑️ F-08 AI Sketches button (the sketch_* API methods are still available via pywebview)
🗑️ F-04 Inspector button + side dock
🗑️ F-01 Timeline dock + toolbar button
🗑️ F-12 Render Farm sidebar toggle + progress modal
🗑️ Narration Voice & Speed sidebar selector (narration itself still works; localStorage-driven)

v1.1.2.0

✨ CLI Mode — same EXE works as headless CLI: ManimStudio render scene.py --quality 1080p --width 1920 --height 1080 --fps 60
✨ MCP Server for Codex — ManimStudio mcp exposes render/validate/presets as MCP tools over stdio; register in ~/.codex/config.toml and Codex can render Manim animations directly
✨ Full Resolution Control — CLI and MCP support all quality presets (120p–8K) plus custom --width/--height override
✨ AI Agent Mode — autonomous loop: generate code → render → capture screenshots → visual review → auto-fix → repeat until correct
✨ Dual Agent Providers — AI Agent works with both Claude Code and OpenAI Codex CLI, using the same provider for edit and review
✨ Continuous Chat — multi-turn conversations with session memory; Claude uses --session-id / --resume for context persistence
✨ Token Optimization — removed code embedding from all prompts (AI reads scene.py from disk), session caching saves 70-80% on repeated turns, screenshots downscaled to 960px for 4x fewer image tokens
✨ Claude Agent Session Reuse — first edit uses --session-id, subsequent edits use --resume so Claude remembers what it already tried
✨ Screenshot Downscaling — review frames auto-resized to 960px wide via Pillow (falls back gracefully if unavailable)
✨ Claude Review via File Read — Claude examines screenshot files from disk using its Read tool instead of base64 prompt embedding
✨ Assets Folder Access — AI workspaces get a junction/symlink to the assets folder; CLAUDE.md/AGENTS.md lists available asset files
✨ Inline Autocomplete — ghost-text code completions powered by Claude Haiku (fast, lightweight)
✨ New Chat Button — reset AI Edit session and start fresh without restarting the panel
✨ Persistent Chat History — AI chat sessions auto-saved to disk (~/.manim_studio/chat_history/); browse, resume, or delete past sessions from the history dropdown
✨ AI-Friendly GUI — aria-label, data-testid, aria-live attributes on all interactive elements; hidden app state indicator (#appStateIndicator) exposes render/AI/file state for computer-control agents (Claude Code Desktop, Codex)
✨ Faster Startup — disk-cached dependency checks (1-hour TTL) skip subprocess calls on repeat launches; Python and LaTeX checks run in parallel threads; shutil.which() fast path for LaTeX
✨ Editor Skeleton UI — CSS shimmer placeholder shown while Monaco editor loads for improved perceived startup speed
🐛 Fixed Codex agent not working (cascading bugs: missing AGENTS.md, wrong CLI flags, no JSONL parsing)
🐛 Fixed Codex agent review using Claude instead of Codex — now each provider reviews with its own CLI
🐛 Fixed --skip-git-repo-check incorrectly added to Claude CLI calls (only valid for Codex)
🐛 Fixed --image flag used for Claude CLI (doesn't exist) — Claude now reads image files via Read tool
🐛 Fixed --output-format stream-json requiring --verbose flag for Claude agent edits
🐛 Fixed agent loop not continuing when review found bugs (edit failures, unchanged code, review errors)
🐛 Fixed review nitpicking minor issues — now focuses on real visual bugs (overlap, off-screen, wrong text, goal mismatch)
🐛 Fixed agent sending only 5 sampled frames — now sends ALL captured frames to review
🔧 Unified all 3 MD templates (non-agent, Claude agent, Codex agent) with consistent Manim context and "ALWAYS read scene.py FIRST" rule
🔧 Instructions piped via stdin (-p without arg) to avoid OS argument length limits
🔧 Review prompt no longer embeds code — reviewer only needs screenshots + goal description
🔧 Agent edit failure counter prevents infinite loops (stops after 5 consecutive failures)
🔧 Build: console mode changed from disable to attach so the EXE works as both GUI and CLI
🔧 Build: removed --force-stdout-spec/--force-stderr-spec (they broke CLI/MCP stdio)
🔒 Security: Shell injection fix — Codex CLI commands converted from shell=True string interpolation to safe list-based subprocess.Popen (prevents model name / image path injection)
🔒 Security: Path traversal fix — upload_file_content now strips directory components with os.path.basename() (prevents writing files outside assets folder)
🔒 Security: XSS fix in asset display — displayAssets() now escapes file.name and file.path via escapeHtml() before innerHTML insertion
🔒 Security: XSS fix in package manager — PyPI search results and installed package list now escape all metadata (pkg.name, pkg.version, pkg.summary, update warnings) before rendering

v1.1.1.0

✨ Claude Code Integration (Beta) — AI Edit now supports Claude Code via claude -p --output-format stream-json, no API key needed
✨ Enhanced Streaming Output — tool actions (Write, Read, Bash) render as styled blocks with icons, code lines get line-number gutters
✨ AI Edit Beta Badge — panel and window mode show "(Beta)" to indicate active development
✨ Auto Narration (Kokoro TTS) — narrate("text") calls generate TTS audio, auto-merged with video on render
✨ Subtitles (CC) — narrated videos get auto-generated WebVTT subtitles in the preview player
✨ Kokoro model auto-download — ~310MB TTS model downloaded automatically on first use
✨ narrate PyPI package — optional narrate[kokoro] for TTS support
✨ Scene Outline Panel — tree view of classes/methods with click-to-navigate
✨ Command Palette (Ctrl+Shift+P), Shortcuts Modal (Ctrl+/), Zen Mode (F11), Bookmarks (Ctrl+Shift+K)
✨ Drag-and-drop .py files onto the editor to open them
✨ AI Edit Panel — Codex + Claude Code, image upload, web search, model selector
✨ Manim Color Picker, Go-to-Definition (F12), Find References (Shift+F12)
✨ Render History — persistent log with replay, open, and delete
✨ Screenshot Save — save preview frames with native Save As dialog
🐛 Fixed WebSearch permission prompts getting stuck on numbered menu format
🐛 Fixed bundled EXE 404 for ai-edit-window.html (absolute path fix)
🐛 Fixed version mismatch, video preview via HTTP, CLI detection, subprocess piping, ANSI stripping
🔧 Refactored Claude Code from PTY-based to stream-json (removed winpty dependency)
🔧 Replaced hardcoded C:\Windows paths with %SystemRoot%
🔧 Added workspace instruction file (AGENTS.md) for reliable AI file editing

v1.1.0.0

✨ basedpyright IntelliSense — real-time type checking, diagnostics, hover docs, and completions powered by a full LSP server
✨ VS Code-quality signature help — live parameter hints when typing ( or , for all Python builtins and Manim classes
✨ Rich per-parameter documentation — typed labels, descriptions, default values, and code examples for 80+ functions
✨ Missing packages auto-detection — startup check notifies existing users of missing required packages with one-click install
✨ basedpyright added to required packages — automatically installed in new and existing environments
🐛 Fixed autosave messages incorrectly appearing in the terminal box
🐛 Added unsaved changes warning before opening a new file
🐛 Render status feedback ("Rendering...") shown immediately on render start
🔧 LSP workspace uses real on-disk directory so basedpyright resolves Manim imports correctly
🔧 Improved terminal polling rate for smoother output streaming

v1.0.9.0

✨ Auto-open output folder after saving rendered files
✨ Improved render controls sidebar with chevron collapse icon
🎨 Enhanced toast notifications with comprehensive logging
🐛 Fixed async button handler for folder opening
🐛 Improved DPI awareness and responsive scaling
📖 Added comprehensive README documentation

v1.0.8.0

✨ Added Package Manager with visual interface
✨ Drag & drop asset uploads
🐛 Fixed GPU toggle persistence
🐛 Fixed autosave recovery dialog
🎨 Improved responsive scaling and DPI awareness

v1.0.7.0

✨ Added auto-save with backup recovery
✨ Integrated xterm.js terminal
🐛 Fixed LaTeX detection
🎨 Improved dark theme

v1.0.6.0

✨ Initial public release
🎨 Professional UI with Monaco Editor
⚡ Dual render modes (Preview/Final)
📁 Asset management system

Made with ❤️ for the Manim Community

⬆ Back to Top

Name		Name	Last commit message	Last commit date
Latest commit History 262 Commits
prompts		prompts
web		web
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
agent_memory.py		agent_memory.py
ai_edit.py		ai_edit.py
ai_sketch.py		ai_sketch.py
app.py		app.py
build_nuitka.py		build_nuitka.py
cli.py		cli.py
icon.ico		icon.ico
inspector.py		inspector.py
multi_scene.py		multi_scene.py
narration_addon.py		narration_addon.py
privacy_policy.txt		privacy_policy.txt
render_farm.py		render_farm.py
timeline.py		timeline.py
visual_diff.py		visual_diff.py

Folders and files

Latest commit

History

Repository files navigation

🎬 Manim Animation Studio

✨ Features

🎨 Professional Code Editor

🤖 AI Edit Panel (Codex + Claude Code)

🎙️ Auto Narration (Kokoro TTS)

⚡ Dual Render Modes

📁 Integrated Asset Management

📦 Visual Package Manager

🖥️ Full-Featured Terminal

🖥️ CLI & MCP Server (Codex Plugin)

🎯 Smart Workflow Features

🌓 Modern UI/UX

📸 Screenshots

Main Workspace

Asset Manager

Package Manager

🚀 Getting Started

Prerequisites

Installation

Building from Source

Building Executable with Nuitka

CLI Usage

Codex MCP Integration

📖 Quick Start Guide

Creating Your First Animation

Adding Narration

Keyboard Shortcuts

🎨 Using Assets

Adding Custom Fonts

Adding Images/Videos

Adding Audio

⚙️ Configuration

Render Settings

GPU Acceleration

Editor Settings

📂 File Structure

🐛 Troubleshooting

Rendering Fails

Custom Fonts Not Working

LaTeX Not Working

IntelliSense Not Working

Package Installation Fails

Video Won't Play in Preview

🤝 Contributing

Development Setup

📜 License

🙏 Acknowledgments

📞 Support

🗺️ Roadmap

Upcoming Features

📊 System Requirements

Minimum

Recommended

🎓 Learning Resources

Manim Tutorials

Python Resources

⭐ Star History

📝 Changelog

v1.1.3.0 (Latest)

v1.1.2.0

v1.1.1.0

v1.1.0.0

v1.0.9.0

v1.0.8.0

v1.0.7.0

v1.0.6.0

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 1

Packages 0

Uh oh!

Packages