AEM MCP Server (aem-mcp-server)
AEM MCP Server is a full-featured, extensible Model Context Protocol (MCP) server for Adobe Experience Manager (AEM). It provides a robust REST/JSON-RPC API for content, component, and asset management, with advanced integrations for AI, chatbots, and automation. This project is designed for AEM developers, content teams, and automation engineers who want to manage AEM programmatically or via natural language.
Table of Contents
Overview
- Modern, TypeScript-based AEM MCP server
- REST/JSON-RPC API for AEM content, component, and asset operations
- AI/LLM integration (OpenAI, Anthropic, Ollama, custom HTTP APIs)
- Telegram bot for conversational AEM management
- Production-ready, modular, and extensible
Features
- AEM Page & Asset Management: Create, update, delete, activate, deactivate, and replicate pages and assets
- Component Operations: Validate, update, scan, and manage AEM components (including Experience Fragments)
- Advanced Search: QueryBuilder, fulltext, fuzzy, and enhanced page search
- Replication & Rollout: Publish/unpublish content, roll out changes to language copies
- Text & Image Extraction: Extract all text and images from pages, including fragments
- Template & Structure Discovery: List templates, analyze page/component structure
- JCR Node Access: Legacy and modern node/content access
- AI/LLM Integration: Natural language interface for AEM via OpenAI, Anthropic, Ollama, or custom LLMs
- Telegram Bot: Manage AEM via chat, including conversational commands
- Security: Auth, environment-based config, and safe operation defaults
Quick Start
Prerequisites
- Node.js 18+
- Access to an AEM instance (local or remote)
Installation
Build
Run (Production)
Run (Development, hot reload)
Usage Examples
1. List all pages under a path
curl -u admin:admin \
-X POST http://localhost:8080/api \
-H 'Content-Type: application/json' \
-d '{"method":"mcp_aem-mcp_listPages","params":{"siteRoot":"/content/we-retail","depth":2,"limit":10}}'
2. Update a component property
curl -u admin:admin \
-X POST http://localhost:8080/api \
-H 'Content-Type: application/json' \
-d '{"method":"mcp_aem-mcp_updateComponent","params":{"componentPath":"/content/we-retail/us/en/experience/jcr:content/root/hero_image","properties":{"Heading":"New Heading"}}}'
3. Use with AI IDEs (Cursor, Cline, etc.)
Configuration
Environment Variables
Create a .env file in the project root with the following (edit as needed):
AEM_HOST=http://localhost:4502
AEM_SERVICE_USER=admin
AEM_SERVICE_PASSWORD=admin
MCP_PORT=8080
GATEWAY_PORT=3000
OPENAI_API_KEY=your-openai-key
TELEGRAM_BOT_TOKEN=your-telegram-bot-token
MCP_USERNAME=admin
MCP_PASSWORD=admin
MCP Client Configuration
Sample for AI-based code editors or custom clients:
{
"mcpServers": {
"aem-mcp": {
"command": "node",
"args": [
"absolute path to dist/mcp-server.js"
]
}
}
}
API & Client Usage
- REST/JSON-RPC: Exposes all AEM operations via HTTP endpoints
- Supported Operations: Page/asset CRUD, component validation/update, search, rollout, publish, text/image extraction, and more
- AI/LLM: Send natural language commands to the server (via API or Telegram)
- Telegram Bot: Connect your bot using
TELEGRAM_BOT_TOKEN and chat with your AEM instance
AI IDE Integration (Cursor, Cline, etc.)
AEM MCP Server is compatible with modern AI IDEs and code editors that support MCP protocol, such as Cursor and Cline.
How to Connect:
- Install and run the AEM MCP Server as described above.
- Configure your IDE to connect to the MCP server. Example for Cursor/Cline:
- Open your IDE's MCP server settings.
- Add a new server with:
- Type: Custom MCP
- Command:
node - Args:
["/absolute/path/to/dist/mcp-server.js"] - Port:
8080 (or as configured) - Auth: Use
MCP_USERNAME/MCP_PASSWORD from your .env
- Restart your IDE and connect. The IDE will now be able to:
- List, search, and manage AEM content
- Run MCP methods (CRUD, search, rollout, etc.)
- Use AI/LLM features if enabled
Custom MCP Clients
- You can build your own MCP client in any language that supports HTTP/JSON-RPC.
- See the Usage Examples for API call patterns.
- Authenticate using basic auth (
MCP_USERNAME/MCP_PASSWORD). - All MCP methods are available via the
/api endpoint.
Security
- Auth required for all operations (see
MCP_USERNAME/MCP_PASSWORD) - Environment-based configuration for safe deployment
- All destructive operations require explicit parameters and validation
Project Structure
src/ — TypeScript source codedist/ — Compiled JS output
Integrations
- AI/LLM: OpenAI, Anthropic, Ollama, custom HTTP APIs
- Telegram: Chat-based AEM management
Contribution
Contributions are welcome! Please open issues or pull requests for bug fixes, features, or documentation improvements.
License
MIT