AbraFlexi MCP Server
Provides a comprehensive MCP interface to the AbraFlexi ERP REST API, enabling management of invoices, contacts, products, bank transactions, and other evidence through natural language.
README
AbraFlexi MCP Server
A comprehensive Model Context Protocol (MCP) server for AbraFlexi integration using FastMCP and python-abraflexi. This server provides complete access to AbraFlexi REST API functionality through MCP-compatible tools.
Features
📄 Invoice Management
invoice_issued_get- Retrieve issued invoices (faktura-vydana)invoice_issued_create- Create new issued invoicesinvoice_issued_update- Update existing issued invoicesinvoice_issued_delete- Remove issued invoicesinvoice_received_get- Retrieve received invoices (faktura-prijata)invoice_received_create- Create new received invoices
👥 Contact Management
contact_get- Retrieve contacts and companies (adresar)contact_create- Create new contactscontact_update- Update existing contactscontact_delete- Remove contacts
📦 Product Management
product_get- Retrieve products from price list (cenik)product_create- Create new productsproduct_update- Update existing productsproduct_delete- Remove products
🏦 Bank Transaction Management
bank_transaction_get- Retrieve bank transactions (banka)bank_transaction_create- Create new bank transactions
🔧 Generic Evidence Operations
evidence_get- Get records from any evidenceevidence_create- Create record in any evidenceevidence_update- Update record in any evidenceevidence_delete- Delete record from any evidenceevidence_list- List all available evidences
Installation
Prerequisites
- Python 3.10 or higher
- Access to an AbraFlexi server with API enabled
Option 1: Install from PyPI (Recommended)
pip install abraflexi-mcp-server
Then run the server:
abraflexi-mcp
Option 2: AppImage (Linux)
Download the self-contained AppImage from the latest release:
chmod +x AbraFlexi-MCP-Server-*-x86_64.AppImage
./AbraFlexi-MCP-Server-*-x86_64.AppImage
No Python or pip required. See AppImage section for details.
Option 3: Install from Source
-
Clone the repository:
git clone https://github.com/VitexSoftware/abraflexi-mcp-server.git cd abraflexi-mcp-server -
Install with uv (recommended):
uv sync uv run python scripts/start_server.pyOr with pip:
pip install -e . abraflexi-mcp
Cloud Deployment
A testing deployment is available at:
🌐 https://abraflexi.fastmcp.app/mcp
This cloud-hosted instance allows you to test and use the AbraFlexi MCP server without local installation. Configure your MCP client to connect to this endpoint with HTTP transport.
Note: This is a testing deployment. For production use, we recommend self-hosting using one of the installation methods above.
Configuration
Create a .env file or set environment variables:
cp .env.example .env
# Edit .env with your AbraFlexi server details
Configuration
Required Environment Variables
ABRAFLEXI_URL- Your AbraFlexi server URL (e.g.,https://demo.flexibee.eu:5434)ABRAFLEXI_COMPANY- Company identifier (e.g.,demo_de)
Authentication (choose one method)
Method 1: Username/Password (Recommended)
ABRAFLEXI_LOGIN- Your AbraFlexi usernameABRAFLEXI_PASSWORD- Your AbraFlexi password
Method 2: Session ID
ABRAFLEXI_AUTHSESSID- Your AbraFlexi session ID
Optional Configuration
READ_ONLY- Set totrue,1, oryesto enable read-only mode (default:true)ABRAFLEXI_TIMEOUT- Request timeout in seconds (default:300)
Transport Configuration
ABRAFLEXI_MCP_TRANSPORT- Transport type:stdio(default) orstreamable-http
HTTP Transport Configuration (only used when ABRAFLEXI_MCP_TRANSPORT=streamable-http):
ABRAFLEXI_MCP_HOST- Server host (default:127.0.0.1)ABRAFLEXI_MCP_PORT- Server port (default:8000)ABRAFLEXI_MCP_STATELESS_HTTP- Stateless mode (default:false)AUTH_TYPE- Must be set tono-authfor streamable-http transport
Usage
Running the Server
With startup script (recommended):
uv run python scripts/start_server.py
Direct execution:
uv run python -m abraflexi_mcp_server.server
Transport Options
The server supports two transport methods:
STDIO Transport (Default)
Standard input/output transport for MCP clients like Claude Desktop:
# Set in .env or environment
ABRAFLEXI_MCP_TRANSPORT=stdio
HTTP Transport
HTTP-based transport for web integrations:
# Set in .env or environment
ABRAFLEXI_MCP_TRANSPORT=streamable-http
ABRAFLEXI_MCP_HOST=127.0.0.1
ABRAFLEXI_MCP_PORT=8000
ABRAFLEXI_MCP_STATELESS_HTTP=false
AUTH_TYPE=no-auth
Testing
Run test suite:
uv run python scripts/test_server.py
Read-Only Mode
When READ_ONLY=true (default), the server will only expose GET operations (retrieve data) and block all create, update, and delete operations. This is useful for:
- 📊 Monitoring dashboards
- 🔍 Read-only integrations
- 🔒 Security-conscious environments
- 🛡️ Preventing accidental modifications
To enable write operations, set READ_ONLY=false in your .env file.
Example Tool Calls
Get all issued invoices:
invoice_issued_get(limit=10)
Get specific invoice by code:
invoice_issued_get(kod="INV-2024-001")
Create a new contact:
contact_create(
kod="CUSTOMER01",
nazev="Example Company s.r.o.",
email="info@example.com",
tel="+420123456789"
)
Get products:
product_get(nazev="Widget", limit=5)
Generic evidence query:
evidence_get(
evidence="faktura-vydana",
filter_expr="datVyst >= '2024-01-01'",
limit=20
)
MCP Integration
This server is designed to work with MCP-compatible clients like Claude Desktop. See MCP_SETUP.md for detailed integration instructions.
OCI Container
The server can be run as an OCI container (Docker/Podman) — no Python installation needed on the host.
Building the image
podman build -t abraflexi-mcp-server -f Containerfile .
Running the container
The image defaults to streamable-http transport on port 8000.
With individual environment variables:
podman run --rm -p 8000:8000 \
-e ABRAFLEXI_URL=https://demo.flexibee.eu:5434 \
-e ABRAFLEXI_COMPANY=demo_de \
-e ABRAFLEXI_LOGIN=winstrom \
-e ABRAFLEXI_PASSWORD=winstrom \
abraflexi-mcp-server
With an env file:
podman run --rm -p 8000:8000 --env-file .env abraflexi-mcp-server
Container environment defaults
| Variable | Default |
|---|---|
ABRAFLEXI_MCP_TRANSPORT |
streamable-http |
ABRAFLEXI_MCP_HOST |
0.0.0.0 |
ABRAFLEXI_MCP_PORT |
8000 |
READ_ONLY |
true |
All other configuration variables can be passed as environment variables.
AppImage
A self-contained, single-file Linux executable — no Python, pip, or any other dependency required on the host.
Building the AppImage
bash appimage/build-appimage.sh
The script downloads a portable CPython and appimagetool automatically. The resulting file is placed in build/appimage/:
build/appimage/AbraFlexi-MCP-Server-<version>-x86_64.AppImage
Running the AppImage
The AppImage automatically loads a .env file from the current working directory if one is present.
With a .env file (recommended):
cp .env.example .env
# edit .env with your credentials
./AbraFlexi-MCP-Server-*-x86_64.AppImage
With inline environment variables:
ABRAFLEXI_URL=https://demo.flexibee.eu:5434 \
ABRAFLEXI_COMPANY=demo_de \
ABRAFLEXI_LOGIN=winstrom \
ABRAFLEXI_PASSWORD=winstrom \
./AbraFlexi-MCP-Server-*-x86_64.AppImage
Development
Project Structure
abraflexi-mcp-server/
├── abraflexi_mcp_server/
│ ├── __init__.py
│ └── server.py # Main server implementation
├── appimage/
│ ├── AppRun # AppImage entry point
│ ├── abraflexi-mcp-server.desktop
│ ├── abraflexi-mcp-server.svg
│ └── build-appimage.sh # AppImage build script
├── debian/ # Debian packaging
├── scripts/
│ ├── start_server.py # Startup script with validation
│ └── test_server.py # Test script
├── Containerfile # OCI container build
├── server.json # MCP Registry manifest
├── pyproject.toml # Python project configuration
├── setup.py # Legacy setuptools configuration
├── requirements.txt # Dependencies
├── .env.example # Environment configuration template
├── .env # Your configuration (not in git)
├── .gitignore # Git ignore patterns
└── README.md # This file
Running Tests
# Test server functionality
uv run python scripts/test_server.py
# Test with specific environment
ABRAFLEXI_URL=https://your-server.com uv run python scripts/test_server.py
Error Handling
The server includes comprehensive error handling:
- ✅ Authentication errors are clearly reported
- 🔒 Read-only mode violations are blocked with descriptive messages
- ✔️ Invalid parameters are validated
- 🌐 Network and API errors are properly formatted
- 📝 Detailed logging for troubleshooting
Security Considerations
- 🔑 Store credentials securely in
.envfile (never commit to git) - 🔒 Enable read-only mode for monitoring-only use cases
- 🛡️ Use HTTPS for AbraFlexi server connections
- 🔄 Regularly rotate passwords
- 📁 Ensure
.envfile has proper permissions (600)
Troubleshooting
Common Issues
Connection Failed:
- Verify
ABRAFLEXI_URLis correct and accessible - Check authentication credentials
- Ensure AbraFlexi API is enabled
- Check firewall/network settings
Permission Denied:
- Verify user has sufficient AbraFlexi permissions
- Check if read-only mode is enabled when trying to modify data
Tool Not Found:
- Ensure all dependencies are installed:
uv sync - Verify Python version compatibility (3.10+)
Debug Mode
Set environment variable for detailed logging:
export DEBUG=1
uv run python scripts/start_server.py
Dependencies
- FastMCP - MCP server framework
- python-abraflexi - AbraFlexi Python library
- python-dotenv - Environment variable management
License
This project is licensed under the MIT License.
Acknowledgments
- AbraFlexi for the accounting platform
- Model Context Protocol for the integration standard
- FastMCP for the server framework
Support
Author
Vítězslav Dvořák
- Email: info@vitexsoftware.cz
- GitHub: @VitexSoftware
Made with ❤️ for the AbraFlexi and MCP communities
<!-- mcp-name: io.github.Vitexus/abraflexi -->
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。