ERPNext MCP Server
This MCP server connects ChatGPT to ERPNext, enabling AI-powered business analytics, report generation, and document management.
README
ERPNext MCP Server for ChatGPT
A complete Model Context Protocol (MCP) integration that connects ChatGPT to your ERPNext instance, enabling AI-powered business analytics, report generation, and document management.
What This Does
This integration allows ChatGPT to:
- Generate financial reports (P&L, Balance Sheet, Cash Flow)
- Analyze sales trends and customer data
- Check inventory and stock levels
- Access HR and employee information
- Create and update ERPNext documents
- Build custom dashboards and visualizations
Project Structure
/home/frappe/test.bench/apps/erp_chatbot/
├── erp_chatbot/
│ ├── api.py # Main API endpoints
│ ├── mcp_server_sse.py # Standalone MCP server with SSE transport
│ └── hooks.py # Frappe app hooks
├── chatgpt_openapi.yaml # OpenAPI schema for ChatGPT Actions
├── chatgpt_gpt_instructions.md # GPT behavior instructions
├── QUICK_START.md # 5-minute setup guide
├── API_KEY_SETUP.md # Authentication troubleshooting
└── pyproject.toml # Dependencies
Quick Start (5 Minutes)
Step 1: Install Dependencies
cd /home/frappe/test.bench
./env/bin/pip install mcp sse-starlette starlette uvicorn
Step 2: Get ERPNext API Key
- Login to ERPNext as Administrator
- Go to User → API Access
- Click Generate Keys
- Copy API Key and API Secret
Step 3: Create ChatGPT GPT
- Go to chat.openai.com
- Click Explore GPTs → Create
- Name:
ERPNext Assistant
Step 4: Configure GPT
Instructions Tab:
- Copy contents from
chatgpt_gpt_instructions.md - Paste in Instructions box
Actions Tab:
- Click Add actions
- Import schema:
chatgpt_openapi.yaml - Authentication:
- Type: API Key
- Key:
Authorization - Value:
Bearer YOUR_API_KEY:YOUR_API_SECRET - Auth Type: Bearer
Step 5: Test
Ask your GPT:
"What companies are available?"
"Show me this month's sales"
Available API Endpoints
Public Endpoints (No Auth Required)
| Endpoint | Method | Description |
|---|---|---|
/api/method/erp_chatbot.api.mcp_chatgpt_tools |
GET | List all available tools |
Authenticated Endpoints
| Endpoint | Method | Description |
|---|---|---|
/api/method/erp_chatbot.api.mcp_chatgpt_execute |
POST | Execute MCP tools |
/api/method/erp_chatbot.api.get_financial_analysis |
GET | P&L and financial reports |
/api/method/erp_chatbot.api.get_sales_dashboard |
GET | Sales analytics dashboard |
/api/method/erp_chatbot.api.get_inventory_summary |
GET | Inventory overview |
/api/method/erp_chatbot.api.generate_chart |
POST | Create data visualizations |
/api/method/erp_chatbot.api.create_report |
POST | Build custom reports |
/api/method/erp_chatbot.api.get_dashboard_kpis |
GET | Key performance indicators |
Available Tools
Reporting Tools
get_report_data- Execute standard ERPNext reportsget_sales_analytics- Sales trends with date filteringget_financial_analysis- P&L statements with charts
Data Query Tools
get_list_data- Query any DocType with filtersget_document- Fetch single documentget_companies- List available companiesget_inventory_summary- Stock levels and overview
Document Management
create_document- Create new ERPNext recordsupdate_document- Modify existing records
Analytics & Visualization
generate_chart- Create bar, pie, line chartscreate_report- Build custom reportsget_dashboard_kpis- Business KPIs
Authentication
API Key Format
Authorization: Bearer api_key:api_secret
Example:
Authorization: Bearer d4e8f2c1b5a9:8f3e2d1c9b7a5f4e
Getting API Keys
- ERPNext → Users → Select your user
- Scroll to API Access section
- Click Generate Keys
- Save both API Key and API Secret
Important: API Secret is shown only once. Save it immediately!
Standalone MCP Server
Run the MCP server independently:
cd /home/frappe/test.bench
./env/bin/python apps/erp_chatbot/erp_chatbot/mcp_server_sse.py
Server starts on http://localhost:8000
Endpoints
GET /mcp/sse- SSE connectionPOST /mcp/messages- MCP messagesGET /health- Health check
Usage Examples
Query Sales Data
{
"tool": "get_list_data",
"arguments": {
"doctype": "Sales Invoice",
"filters": {"status": "Overdue"},
"fields": ["name", "customer", "grand_total"],
"limit": 20
}
}
Create a Customer
{
"tool": "create_document",
"arguments": {
"doctype": "Customer",
"data": {
"customer_name": "New Customer Ltd",
"customer_type": "Company",
"territory": "Pakistan"
}
}
}
Generate P&L Report
{
"tool": "get_report_data",
"arguments": {
"report_name": "Profit and Loss Statement",
"filters": {
"company": "Your Company",
"fiscal_year": "2024-2025"
}
}
}
Create a Chart
{
"chart_type": "bar",
"labels": ["Jan", "Feb", "Mar", "Apr"],
"values": [10000, 15000, 12000, 18000],
"title": "Monthly Sales"
}
Troubleshooting
401 Authentication Error
- Verify API Key format:
Bearer key:secret(with colon) - Check user has proper permissions
- Ensure API keys are generated correctly
Tools Not Showing
- Refresh schema in ChatGPT Actions
- Check
chatgpt_openapi.yamlis properly uploaded - Verify server URL is correct
Connection Failed
- Check ERPNext server is running
- Verify domain URL in schema
- Test endpoint directly in browser
Support
For issues:
- Check Frappe logs:
bench --site yoursite.com logs - Verify API permissions
- Test endpoints directly:
https://yoursite.com/api/method/erp_chatbot.api.mcp_chatgpt_tools
Files Reference
| File | Purpose |
|---|---|
chatgpt_openapi.yaml |
OpenAPI schema for ChatGPT |
chatgpt_gpt_instructions.md |
GPT behavior configuration |
api.py |
Backend API endpoints |
mcp_server_sse.py |
Standalone MCP server |
QUICK_START.md |
Quick setup guide |
API_KEY_SETUP.md |
Auth troubleshooting |
Security Notes
- Keep API keys secret and secure
- Use HTTPS in production
- Regularly rotate API keys
- Grant minimum required permissions
Ready to Use!
Your ChatGPT is now connected to ERPNext. Start asking business questions!
Example queries:
- "Show me overdue invoices"
- "What's our inventory status?"
- "Create a sales order for ABC Corp"
- "Generate P&L for Q1 2024"
Built for test.bench / erp.site.com
Created by Ibni for ERPNext MCP Server
License
mit
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。