Microsoft Business Central MCP Server
Model Context Protocol (MCP) server for Microsoft Dynamics 365 Business Central. Provides AI assistants with direct access to Business Central data through properly formatted API v2.0 calls.
README
Microsoft Business Central MCP Server
<img width="1536" height="1024" alt="mcp-business-central" src="https://github.com/user-attachments/assets/13932bfd-a5b9-4668-a7cd-ac9549a09673" />
Model Context Protocol (MCP) server for Microsoft Dynamics 365 Business Central. Provides AI assistants with direct access to Business Central data through properly formatted API v2.0 calls.
Features
- ✅ Correct API URLs: Uses proper
/companies(id)/resourceformat (no ODataV4 segment) - ✅ Zero Installation: Run with
npx- no pre-installation required - ✅ Azure CLI Auth: Leverages existing Azure CLI authentication
- ✅ Client Credentials Auth: Service-to-service authentication for AI agents
- ✅ Clean Tool Names: No prefixes, just
get_schema,list_items, etc. - ✅ Full CRUD: Create, read, update, and delete Business Central records
Installation
Using npx (Recommended)
No installation needed! Configure in Claude Desktop or Claude Code:
{
"mcpServers": {
"business-central": {
"type": "stdio",
"command": "cmd",
"args": ["/c", "npx", "-y", "@knowall-ai/mcp-business-central"],
"env": {
"BC_URL_SERVER": "https://api.businesscentral.dynamics.com/v2.0/{tenant-id}/{environment}/api/v2.0",
"BC_COMPANY": "Your Company Name",
"BC_AUTH_TYPE": "azure_cli"
}
}
}
}
Note for Windows: Use cmd with /c as shown above for proper npx execution.
Using Smithery
Install via Smithery:
npx -y @smithery/cli install @knowall-ai/mcp-business-central --client claude
Local Development
git clone https://github.com/knowall-ai/mcp-business-central.git
cd mcp-business-central
npm install
npm run build
node build/index.js
Configuration
Environment Variables
| Variable | Required | Description | Example |
|---|---|---|---|
BC_URL_SERVER |
Yes | Business Central API base URL | https://api.businesscentral.dynamics.com/v2.0/{tenant}/Production/api/v2.0 |
BC_COMPANY |
Yes | Company display name | KnowAll Ltd |
BC_AUTH_TYPE |
No | Authentication type (default: azure_cli) |
azure_cli or client_credentials |
BC_TENANT_ID |
For client_credentials | Azure AD tenant ID | 00000000-0000-0000-0000-000000000000 |
BC_CLIENT_ID |
For client_credentials | App registration client ID | 00000000-0000-0000-0000-000000000000 |
BC_CLIENT_SECRET |
For client_credentials | App registration client secret | your-secret-value |
Getting Your Configuration Values
- Tenant ID: Find in Azure Portal → Azure Active Directory → Overview
- Environment: Usually
ProductionorSandbox - Company Name: The display name shown in Business Central
Example URL format:
https://api.businesscentral.dynamics.com/v2.0/00000000-0000-0000-0000-000000000000/Production/api/v2.0
Authentication
Recommendation: Use
azure_cliauthentication - it's simpler to set up and more reliable. Theclient_credentialsmethod is also supported but has known configuration challenges with Business Central's Microsoft Entra Applications setup. See docs/TROUBLESHOOTING.adoc for details.
Option 1: Azure CLI (Recommended)
The simplest and most reliable authentication method. Uses your existing Azure CLI login.
Prerequisites:
- Install Azure CLI: https://docs.microsoft.com/cli/azure/install-azure-cli
- Login:
az login - Verify access:
az account get-access-token --resource https://api.businesscentral.dynamics.com
Configuration:
{
"mcpServers": {
"business-central": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@knowall-ai/mcp-business-central"],
"env": {
"BC_AUTH_TYPE": "azure_cli",
"BC_URL_SERVER": "https://api.businesscentral.dynamics.com/v2.0/{tenant-id}/Production/api/v2.0",
"BC_COMPANY": "My Company"
}
}
}
}
Option 2: Client Credentials (Service-to-Service)
For automated systems that need to run without user interaction. This method uses OAuth 2.0 client credentials flow.
Note: This method has known configuration challenges. The Business Central "Microsoft Entra Applications" setup can be complex and the application user creation may not work as expected. See docs/TROUBLESHOOTING.adoc for detailed guidance.
Setup Overview:
-
Create Azure App Registration:
- Go to Azure Portal → Azure Active Directory → App registrations
- Create new registration (single tenant)
- Add API permission: Dynamics 365 Business Central →
app_access(Application permission, NOT Delegated) - Grant admin consent for the permission
- Add redirect URI:
https://businesscentral.dynamics.com/OAuthLanding.htm
-
Generate Client Secret:
- In your app registration, go to Certificates & secrets
- Create a new client secret and save it securely
-
Configure Business Central:
- In Business Central, search for "Microsoft Entra Applications"
- Click + New and enter your app's Client ID
- Set a Description (this becomes the application user name)
- Set State to "Enabled" - you should see "A user named '[Description]' will be created"
- Add permission sets:
D365 BUS FULL ACCESS(recommended) orD365 READ - Leave Company field blank for all companies access
- Click "Grant Consent"
-
Verify Setup:
- The application user should appear in the Users list in Business Central
- If not, see docs/TROUBLESHOOTING.adoc for solutions
References:
Available Tools
1. get_schema
Get OData metadata for a Business Central resource.
Parameters:
resource(string, required): Resource name (e.g.,customers,contacts,salesOpportunities)
Example:
{
"resource": "customers"
}
2. list_items
List items with optional filtering and pagination.
Parameters:
resource(string, required): Resource namefilter(string, optional): OData filter expressiontop(number, optional): Maximum number of items to returnskip(number, optional): Number of items to skip for pagination
Example:
{
"resource": "customers",
"filter": "displayName eq 'Contoso'",
"top": 10
}
3. get_items_by_field
Get items matching a specific field value.
Parameters:
resource(string, required): Resource namefield(string, required): Field name to filter byvalue(string, required): Value to match
Example:
{
"resource": "contacts",
"field": "companyName",
"value": "Contoso Ltd"
}
4. create_item
Create a new item in Business Central.
Parameters:
resource(string, required): Resource nameitem_data(object, required): Item data to create
Example:
{
"resource": "contacts",
"item_data": {
"displayName": "John Doe",
"companyName": "Contoso Ltd",
"email": "john.doe@contoso.com"
}
}
5. update_item
Update an existing item.
Parameters:
resource(string, required): Resource nameitem_id(string, required): Item ID (GUID)item_data(object, required): Fields to update
Example:
{
"resource": "customers",
"item_id": "1366066e-7688-f011-b9d1-6045bde9b95f",
"item_data": {
"displayName": "Updated Name"
}
}
6. delete_item
Delete an item from Business Central.
Parameters:
resource(string, required): Resource nameitem_id(string, required): Item ID (GUID)
Example:
{
"resource": "contacts",
"item_id": "a1b2c3d4-e5f6-g7h8-i9j0-k1l2m3n4o5p6"
}
Common Resources
companies- Company informationcustomers- Customer recordscontacts- Contact recordssalesOpportunities- Sales opportunitiessalesQuotes- Sales quotessalesOrders- Sales orderssalesInvoices- Sales invoicesitems- Product/service itemsvendors- Vendor records
Troubleshooting
See docs/TROUBLESHOOTING.adoc for detailed troubleshooting guides covering:
- Authentication issues (401 errors, token problems)
client_credentialssetup challenges and known issues- Company not found errors
- Environment-specific configuration (Production vs Sandbox)
Development
# Install dependencies
npm install
# Build TypeScript
npm run build
# Watch mode for development
npm run dev
License
MIT
Contributing
Issues and pull requests welcome at https://github.com/knowall-ai/mcp-business-central
Related Projects
- MCP Specification: modelcontextprotocol.io
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。