SAP OData to MCP Server for BTP

by lemaiwo

Node.js MCP server for SAP OData services integration enabling conversational AI interface with SAP S/4HANA or ECC systems via MCP protocol.

Business & enterprisestdioCommunity

Repository-wide counts · Cached 2026-05-22

Overview

The SAP OData to MCP Server for BTP MCP server is a publicly available project. Review the upstream repository for installation instructions, supported tools, compatibility, permissions, and current maintenance status.

Configuration

Configuration, transport, authentication, and runtime requirements vary by project. Open the repository before connecting and use the smallest set of credentials and permissions required.

Open the SAP OData to MCP Server for BTP repository to read the latest documentation.

KEEP EXPLORING

Compare source, connection, and authentication details before choosing an implementation.

View the complete category

MCP Atlassian

sooperset

Community

Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Supports Cloud and Server/Data Center deployments. Distributed as a Docker image. Configure via environment variables for authentication and filtering.

Google Workspace MCP Server

Taylor Wilsdon

Community

一款生产就绪的 MCP 服务器,集成了 Google Workspace 服务(日历、云端硬盘、Gmail、文档、表格、幻灯片、表单、聊天)与 AI 助手。需要在项目根目录的 client_secret.json 文件中配置 Google Cloud OAuth 2.0 凭据,并设置环境变量以指定服务器的基础 URI 和端口。

Meta Ads MCP

nictuku

Community

Model Context Protocol (MCP) server for interacting with Meta Ads API. Enables AI models to access, analyze, and manage Meta advertising campaigns through a standardized interface. Requires PIPEBOARD_API_TOKEN environment variable for authentication with Pipeboard service.

Google 广告 MCP

cohnen

Community

用于模型上下文协议(MCP)的 Google 广告 API 集成。需要通过环境变量或 JSON 配置文件(详见 README)配置 Google 广告 API 凭据。

FROM THE SOURCE

Repository README

Build-time snapshot · Retrieved 2026-10-05

View original

SAP OData to MCP Server for BTP🚀

🎯 Project Goal

Transform your SAP S/4HANA or ECC system into a conversational AI interface by exposing all OData services as dynamic MCP tools. This enables natural language interactions with your ERP data:

  • "Show me 10 banks" → Automatically queries the Bank entity with $top=10
  • "Update bank with ID 1 to have street number 5" → Executes PATCH operation on Bank entity
  • "Create a new customer with name John Doe" → Performs POST to Customer entity
  • "List all purchase orders from this week" → Applies $filter for date range on PurchaseOrder entity

🏗️ Architecture Overview - 3-Level Progressive Discovery

graph TB
    A[AI Agent/LLM] --> B[MCP Client]
    B --> C[SAP MCP Server]
    C --> D[SAP BTP Destination]
    D --> E[SAP System]

    C --> F[Level 1: Lightweight Discovery]
    F --> G[Minimal Service/Entity List]
    C --> H[Level 2: Full Metadata]
    H --> I[Complete Entity Schemas]
    C --> J[Level 3: CRUD Execution]
    J --> K[Authenticated Operations]

    style A fill:#e1f5fe
    style C fill:#f3e5f5
    style E fill:#e8f5e8
    style F fill:#fff3e0
    style H fill:#e8eaf6
    style J fill:#e0f2f1

Core Components:

  1. 🔍 Level 1 - Discovery: Lightweight search returning minimal service/entity lists (token-optimized)
  2. 📋 Level 2 - Metadata: Full schema details on-demand for selected entities
  3. ⚡ Level 3 - Execution: Authenticated CRUD operations using metadata from Level 2
  4. 🔌 MCP Protocol Layer: Full compliance with MCP 2025-06-18 specification
  5. 🌐 HTTP Transport: Session-based Streamable HTTP for web applications
  6. 🔐 BTP Integration: Seamless authentication via SAP BTP Destination service

3-Level Approach Benefits:

  • Token Efficient: Level 1 returns 90% less data than full schemas
  • Progressive Detail: Fetch full schemas only when needed
  • Better LLM Experience: Smaller responses, clearer workflow
  • Reduced Context: From 200+ tools down to just 3

✨ Key Features

🎨 Natural Language to OData

  • Smart Query Translation: Converts natural language to proper OData queries
  • Context-Aware Operations: Understands entity relationships and constraints
  • Parameter Inference: Automatically maps user intent to tool parameters

🔄 Dynamic CRUD Operations

  • Read Operations: Entity sets with filtering, sorting, pagination
  • Create Operations: New entity creation with validation
  • Update Operations: Partial and full entity updates
  • Delete Operations: Safe entity deletion with confirmation

🚀 Production-Ready

  • Session Management: Automatic session creation and cleanup
  • Error Handling: Comprehensive error handling with user-friendly messages
  • Logging: Detailed logging for debugging and monitoring
  • Security: DNS rebinding protection, CORS, Helmet security

📊 Real-Time Metadata

  • Service Catalog: Live discovery of available services
  • Entity Schemas: Dynamic schema generation from OData metadata
  • Capability Detection: Automatic detection of CRUD capabilities per entity

🏛️ System Architecture

┌─────────────────────┐    ┌───────────────────────────┐    ┌─────────────────────┐
│                     │    │                           │    │                     │
│   🤖 AI Agent       │    │   🖥️  SAP MCP Server     │    │   🏢 SAP            │
│   - Claude          │◄──►│   - Service Discovery     │◄──►│   - OData Services  │
│   - GPT-4           │    │   - CRUD Tool Registry    │    │   - Business Logic  │
│   - Local LLMs      │    │   - Session Management    │    │   - Master Data     │
│                     │    │   - BTP Authentication    │    │                     │
└─────────────────────┘    └───────────────────────────┘    └─────────────────────┘
                                           │                                       
                                           ▼                                       
                           ┌───────────────────────────┐                          
                           │                           │                          
                           │   ☁️  SAP BTP Platform    │                          
                           │   - Destination Service   │                          
                           │   - Connectivity Service  │                          
                           │   - XSUAA Security        │                          
                           │                           │                          
                           └───────────────────────────┘                          

🎯 Use Cases

📈 Business Intelligence Queries

User: "Show me top 10 customers by revenue this quarter"
→ Tool: r-CustomerService-Customer
→ Parameters: $filter, $orderby, $top

📝 Data Maintenance

User: "Update supplier ABC123 to have status 'Active'"
→ Tool: u-SupplierService-Supplier
→ Parameters: SupplierId="ABC123", Status="Active"

📊 Analytical Insights

User: "How many open purchase orders are there?"
→ Tool: r-PurchaseOrderService-PurchaseOrder
→ Parameters: $filter=Status eq 'Open'&$count=true

🔧 System Administration

User: "List all inactive users in the system"
→ Tool: r-UserService-User
→ Parameters: $filter=Status eq 'Inactive'

🛠️ Installation & Setup

Prerequisites

  • Node.js 18.x or higher
  • SAP S/4HANA or ECC system with OData services enabled
  • SAP BTP account with Destination and Connectivity services
  • TypeScript knowledge for customization

🚀 Usage Examples

Natural Language Queries

The MCP server automatically translates these natural language commands to the appropriate tool calls:

Natural Language Generated Tool Call OData Query
"Show me 10 banks" r-BankService-Bank GET /BankSet?$top=10
"Find banks in Germany" r-BankService-Bank GET /BankSet?$filter=Country eq 'DE'
"Update bank 123 name to ABC Corp" u-BankService-Bank PATCH /BankSet('123')
"Create a new customer John Doe" c-CustomerService-Customer POST /CustomerSet
"Delete order 456" d-OrderService-Order DELETE /OrderSet('456')

📋 Available Tools - 3-Level Architecture

The server exposes 3 progressive discovery tools instead of hundreds of individual CRUD tools:

Level 1: discover-sap-data

Purpose: Lightweight search for services and entities

Returns: Minimal data (serviceId, serviceName, entityName, entityCount)

Usage:

// Search for customer entities
discover-sap-data({ query: "customer" })

// Get all available services
discover-sap-data({ query: "" })

// Search in specific category
discover-sap-data({ query: "sales", category: "sales" })

Fallback: If no matches found, returns ALL services with entity lists


Level 2: get-entity-metadata

Purpose: Get complete schema for a specific entity

Returns: Full schema with properties, types, keys, capabilities

Usage:

// Get full schema for Customer entity
get-entity-metadata({
  serviceId: "API_BUSINESS_PARTNER",
  entityName: "Customer"
})

Output: All properties, types, nullable flags, maxLength, keys, capabilities


Level 3: execute-sap-operation

Purpose: Perform authenticated CRUD operations

Operations: read, read-single, create, update, delete

Usage:

// Read customers
execute-sap-operation({
  serviceId: "API_BUSINESS_PARTNER",
  entityName: "Customer",
  operation: "read",
  filterString: "CustomerName eq 'ACME'"
})

// Update customer
execute-sap-operation({
  serviceId: "API_BUSINESS_PARTNER",
  entityName: "Customer",
  operation: "update",
  parameters: { CustomerID: "123", CustomerName: "New Name" }
})

Workflow Example

1. discover-sap-data → "customer"
   ↓ Returns: List of customer-related entities

2. get-entity-metadata → "API_BUSINESS_PARTNER", "Customer"
   ↓ Returns: Full schema with all properties

3. execute-sap-operation → read/create/update/delete
   ✓ Executes operation with proper parameters

Protocol Version: 2025-06-18

Supported Capabilities:

  • ✅ Tools with listChanged notifications
  • ✅ Resources with listChanged notifications
  • ✅ Logging with level control
  • ✅ Session Management for HTTP transport
  • ✅ Error Handling with proper error codes

Transport Support

  • ✅ Streamable HTTP (recommended)
  • ✅ Stdio for command line usage
  • ✅ Session-based with automatic cleanup
  • ✅ DNS Rebinding Protection

🔒 Security & Authentication

SAP BTP Integration

  • Uses BTP Destination service for S/4HANA or ECC authentication
  • Supports Principal Propagation and OAuth2
  • Automatic token refresh and session management
  • Secure credential storage in BTP

HTTP Security

  • Helmet.js security headers
  • CORS protection with configurable origins
  • DNS rebinding attack prevention
  • Request rate limiting (configurable)

Session Security

  • Automatic session expiration (24h default)
  • Secure session ID generation
  • Session cleanup on server restart
  • Memory leak prevention

📚 API Reference

Health Check

GET /health
{
  "status": "healthy",
  "activeSessions": 3,
  "discoveredServices": 25,
  "version": "2.0.0"
}

Server Info

GET /mcp
{
  "name": "btp-sap-odata-to-mcp-server",
  "protocol": { "version": "2025-06-18" },
  "capabilities": { "tools": {}, "resources": {} },
  "features": ["Dynamic service discovery", "CRUD operations"],
  "activeSessions": 3
}

Documentation

GET /docs
{
  "title": "SAP MCP Server API",
  "endpoints": {...},
  "mcpCapabilities": {...},
  "usage": {...}
}

🎬 Demo

See the MCP server in action:

MCP Demo

⚙️ Environment Variable: Disable ReadEntity Tool Registration

To disable registration of the ReadEntity tool for all entities in all services, set the following in your .env file:

DISABLE_READ_ENTITY_TOOL=true

This will prevent registration of the ReadEntity tool for all entities and services.

⚡ Quick Start