Skip to content

fishforks/AE-agent

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

中文文档 | English

AE-Agent

An AI agent platform for After Effects — run any model, any endpoint, fully under your control.

GitHub Stars License: MIT PRs Welcome

Tip

If this project saves you time, please give it a Star — it’s the biggest motivation for us to keep going!

Note

👋 Want to help build it? We need developers, motion designers, and AE power users — whether you contribute code, docs, or just file issues, it all counts. Check the roadmap and contributing guide to get started.


What is AE-Agent?

AE-Agent brings AI-powered automation directly into After Effects through a CEP panel. Instead of being locked to a single cloud provider, you connect it to any OpenAI-compatible API — whether that's Anthropic, a local Ollama instance, or your own self-hosted model.

Ask in natural language. The agent understands your composition, writes and executes the script, and verifies the result visually — all without leaving After Effects.


Features

🤖 Multi-Model Agent Engine

  • Connects to Claude Code or OpenAI Codex CLI running locally on your machine
  • Switch API endpoints freely via the built-in settings panel
  • Supports any OpenAI-compatible base URL (local, cloud, proxy)

🎬 Deep After Effects Integration

  • Reads full composition structure: layers, effects, keyframes, expressions, masks
  • Writes and executes ExtendScript in real time
  • Verifies results with live frame previews before reporting success
  • Checkpoint system: auto-saves project state before each operation

🖼️ Image Generation

  • Generate textures, moodboards, and sprites directly in chat
  • Auto-imports generated images into the project
  • Pluggable image API: swap OpenRouter for any compatible endpoint

⚙️ Flexible Configuration

  • Floating settings panel — no code editing required
  • All keys stored in localStorage, never transmitted to third parties
  • Separate configuration for chat models and image generation

🧩 Third-Party Plugin Support

  • Red Giant suite: Trapcode Particular / Form / Shine / Starglow, Magic Bullet Looks / Colorista, Universe
  • Video Copilot: Saber, Optical Flares, Element 3D (numeric parameters controllable)
  • Any AE effect that exposes properties: Agent can read and modify numeric parameters of any installed plugin
  • Note: features requiring a dedicated UI panel (e.g. E3D Scene Setup) are outside the automation scope

Supported Capabilities

Category What the Agent Can Do
Layer Management Create, rename, reorganize, batch edit layers
Animation Set keyframes, expressions, easing, staggered timing
Effects Apply and configure any AE effect by name
3rd-Party Plugins Control Trapcode, Magic Bullet, Saber and more via numeric params
Text Animate text with per-character control
Shapes Build complex shape layers and path animations
Scripting Generate and run arbitrary ExtendScript
Keying Configure Keylight, Cryptomatte and other keying effects
Tracking Trigger Warp Stabilizer and camera tracker
Images Generate and import AI-created assets

Getting Started

Requirements

  • Windows 10/11 (macOS support planned)
  • After Effects 2022+
  • Node.js 18+
  • One of: Claude Code CLI or OpenAI Codex CLI
# Install Claude Code CLI
npm install -g @anthropic-ai/claude-code

Installation

# Clone the repo
git clone https://github.com/tiansuo-114/AE-agent.git
cd AE-agent

# Run the installer (sets CEP debug mode + copies files)
.\install.ps1

Restart After Effects, then open Window → Extensions → Atom.

Configure Your API

Click the button in the panel, or open DevTools console:

// Chat / coding agent
localStorage.setItem('ATOM_API_KEY', 'your-api-key')
localStorage.setItem('ATOM_CUSTOM_BASE_URL', 'https://your-endpoint.com')

// Image generation
localStorage.setItem('ATOM_IMAGE_API_KEY', 'your-image-key')
localStorage.setItem('ATOM_IMAGE_BASE_URL', 'https://openrouter.ai/api/v1')

Roadmap

The following is on the horizon — contributions welcome!

  • macOS installer — currently Windows only
  • Gemini CLI support — add Google Gemini as a third agent option
  • Premiere Pro port — apply the same CEP + ExtendScript architecture to Premiere
  • Skills library — community-maintained skill packs for common workflows
  • Video frame analysis — extract frames via ffmpeg and feed to vision models
  • Batch mode — process multiple compositions in queue without supervision ✅
  • MCP integration — expose AE operations as Model Context Protocol tools ✅
  • Plugin awareness — deeper support for Trapcode, Red Giant, and other third-party effects ✅
  • Timeline timing control — describe a timing feel and let the agent match it ✅

Community

Join the QQ group to share workflows, ask questions, and connect with other motion designers using AE-Agent.

Group name AE-agent 交流群
Group ID 798447894

QQ Group QR Code

Scan with QQ app to join directly.

WeChat Group

WeChat Group QR Code

Scan with WeChat to join directly.


Contributing

This project is in active development and contributions of all kinds are welcome.

How to contribute

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/my-idea
  3. Make your changes
  4. Open a Pull Request with a clear description

Good first contributions

  • Test on different AE versions and report compatibility
  • Improve the patch scripts to support newer Atom releases
  • Write Skills (markdown files) for common motion design workflows
  • Add macOS support to the install script
  • Document edge cases in the wiki

Found a bug or have an idea?

Open an Issue — all feedback is appreciated.


Architecture

AE-Agent
├── CEP Panel (Svelte/JS)         ← UI running in AE's embedded browser
│   └── main-Bumuo9MN.js          ← Core bundle (patched)
├── ExtendScript Bridge           ← Talks to AE's scripting engine
│   ├── AEKnowledgeExtractorV2    ← Scans composition structure
│   └── Main.jsx                  ← Script runner
└── Agent CLI                     ← Claude Code or Codex (local process)
    └── stdin/stdout JSON stream  ← Communication protocol

License

MIT — see LICENSE

About

AI agent for After Effects — bring your own API key and endpoint

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • JavaScript 89.5%
  • Python 8.1%
  • PowerShell 1.8%
  • HTML 0.6%