After Effects MCP Server
Enables AI assistants to control Adobe After Effects for composition, layer, effect, preset, marker, and audio automation through a bridge panel.
README
After Effects MCP Server
Control Adobe After Effects through MCP using a bridge panel running inside AE. This project is optimized for practical automation workflows: effects, presets, keyframing, markers, and audio-aware tooling.
Highlights
- Full composition and layer automation.
- Deep effect inspection and property editing.
- Advanced effect graph controls (temporal + spatial).
- Preset search/list/apply workflows.
- Marker and audio automation, including waveform-to-marker pipelines.
- Installed effect catalog discovery (
list-available-effects).
Feature Matrix
| Area | Capabilities |
|---|---|
| Composition | Create, inspect project/compositions, get clip frame ranges |
| Layers | Text/shape/solid/adjustment creation, transform/property updates, centering |
| Animation | Layer keyframes, expressions, effect keyframes with graph controls |
| Effects | Apply by name/matchName, list layer effects recursively, edit any effect property, remove effects |
| Presets | Apply .ffx, list/search preset libraries |
| Markers | Add single marker (comp/layer), add markers in bulk |
| Audio | Set channel levels, inspect audio metadata, analyze WAV waveform, detect peaks |
Requirements
- Adobe After Effects
- Node.js 18+
- npm
In After Effects, enable:
Edit -> Preferences -> Scripting & Expressions -> Allow Scripts to Write Files and Access Network
Quick Start
- Clone and install:
git clone https://github.com/TheLlamainator/after-effects-mcp.git
cd after-effects-mcp
npm install
- Build:
npm run build
- Install bridge script:
npm run install-bridge
- Restart After Effects and open:
Window -> mcp-bridge-auto.jsx- Keep this panel open during MCP usage.
MCP Client Config
Use an absolute path to build/index.js.
{
"mcpServers": {
"AfterEffectsMCP": {
"command": "node",
"args": ["<absolute-path-to-repo>/build/index.js"]
}
}
}
Add to VS Code
If you use an MCP-capable VS Code extension, add this server in that extension's MCP server settings.
Use:
- command:
node - args:
["<absolute-path-to-repo>/build/index.js"]
Example snippet many extensions accept:
{
"mcpServers": {
"AfterEffectsMCP": {
"command": "node",
"args": ["C:\\Users\\<you>\\Documents\\Projects\\AEMCP\\build\\index.js"]
}
}
}
Then:
- Restart VS Code.
- Open After Effects and keep
Window -> mcp-bridge-auto.jsxopen. - Call a simple tool like
get-helporrun-scriptwithgetProjectInfo.
Add to Claude Desktop
Edit Claude Desktop config and add the MCP server entry.
Typical Windows config location:
%APPDATA%\\Claude\\claude_desktop_config.json
Example:
{
"mcpServers": {
"AfterEffectsMCP": {
"command": "C:\\Program Files\\nodejs\\node.exe",
"args": [
"C:\\Users\\<you>\\Documents\\Projects\\AEMCP\\build\\index.js"
]
}
}
}
After saving:
- Fully restart Claude Desktop.
- Open AE bridge panel.
- Verify with
tools/listin logs or by calling a known tool.
Add to Claude Code
Configure the same server command/args in your Claude Code MCP configuration.
Use this server definition:
{
"AfterEffectsMCP": {
"command": "node",
"args": ["<absolute-path-to-repo>/build/index.js"]
}
}
Then:
- Restart Claude Code or reload MCP servers.
- Ensure After Effects is open with
mcp-bridge-auto.jsxpanel running. - Test with
get-resultsafter a queued command.
Architecture Graph
flowchart LR
A[AI Client<br/>VS Code / Claude Desktop / Claude Code] --> B[MCP Server<br/>build/index.js]
B --> C[Bridge Files<br/>ae_command.json / ae_mcp_result.json]
C --> D[After Effects Bridge Panel<br/>mcp-bridge-auto.jsx]
D --> E[Adobe After Effects]
E --> D
D --> C
C --> B
B --> A
Command Flow Graph
sequenceDiagram
participant Client as MCP Client
participant Server as MCP Server
participant Bridge as AE Bridge Panel
participant AE as After Effects
Client->>Server: tools/call (example: apply-effect)
Server->>Server: write ae_command.json
Server-->>Client: queued response (or waits)
Bridge->>Bridge: poll command file
Bridge->>AE: execute command
AE-->>Bridge: result/error
Bridge->>Bridge: write ae_mcp_result.json
Client->>Server: get-results
Server-->>Client: final result JSON
Typical Runtime Flow
- Start your MCP client (it starts this server).
- Keep AE bridge panel open.
- Call tools.
- If response says queued, call
get-resultsafter 1-3 seconds.
Note: some AE operations finish slightly after tool timeout windows; get-results usually contains the final state.
Tool Catalog
General
run-scriptget-resultsget-help
Composition and Layer Utilities
create-compositioncreate-adjustment-layercenter-layersget-layer-clip-frames
Effects and Presets
apply-effectadd-any-effectmcp_aftereffects_applyEffectapply-effect-templatelist-layer-effectslist-available-effectsset-effect-propertyset-effect-keyframeremove-effectapply-presetlist-presetssearch-presets
Markers and Audio
add-markeradd-markers-bulkset-audio-levelsget-audio-infoanalyze-audio-waveform
Diagnostics and Helpers
test-animationrun-bridge-testmcp_aftereffects_get_effects_help
Audio to Marker Workflow
get-audio-infoon target layer.- Copy
sourceFilePath. analyze-audio-waveformwith optionalnumPoints.- Convert
peakTimestomarkers[]. add-markers-bulk.
Project Layout
src/index.ts- MCP server and tool definitionssrc/scripts/mcp-bridge-auto.jsx- AE bridge panelinstall-bridge.js- bridge installer
Development
Build:
npm run build
Install bridge:
npm run install-bridge
Run server directly:
node build/index.js
Troubleshooting
Server does not start
- Run
npm run build. - Check MCP logs for startup exceptions (for example duplicate tool registration).
Commands queue but do not complete
- Ensure AE bridge panel is open.
- Confirm AE scripting/network permission is enabled.
- Retry and call
get-resultsafter a short delay.
Results appear stale
- Reopen bridge panel.
- Send a new command and then call
get-results.
Program Files install fails
- Expected without elevated permissions.
- User AppData script paths are usually sufficient.
License
MIT. See LICENSE.
Inspo
https://github.com/Dakkshin/after-effects-mcp
I took the base code from them but added a lot
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。