ezPay E-Invoice MCP Server

ezPay E-Invoice MCP Server

An MCP server that exposes tools for Taiwan's ezPay e-invoice lifecycle, enabling AI assistants to issue, void, and query invoices and allowances through the standard MCP protocol.

Category
Visit Server

README

mcp-ezpay-einvoice

MCP Server for Taiwan's ezPay E-Invoice API

PyPI version Python License: MIT MCP GitHub stars GitHub issues GitHub last commit

Traditional Chinese (繁體中文)


Overview

An MCP (Model Context Protocol) server that exposes 7 AI-callable tools for the full Taiwan e-invoice lifecycle through the ezPay platform. Covers invoice issuance (B2B/B2C), triggering, voiding, allowances (credit notes/refunds), and querying -- all accessible to AI assistants like Claude via the standard MCP protocol.

Part of the Asgard AI Platform open-source ecosystem.

Features

  • issue_invoice -- Issue B2B (business) or B2C (individual) e-invoices with full item detail, tax types, carriers, and donation codes
  • trigger_invoice -- Trigger deferred or scheduled invoices for immediate issuance
  • void_invoice -- Void issued invoices within the allowed reporting period
  • issue_allowance -- Issue allowances (credit notes / partial refunds) against existing invoices
  • trigger_allowance -- Confirm or cancel pending allowances
  • void_allowance -- Void confirmed allowances
  • query_invoice -- Query invoice details by invoice number or order number
  • AES-256-CBC encryption -- Implements the ezPay encrypted form POST protocol (URL-encode, PKCS7 pad, AES encrypt, hex encode)
  • CheckCode verification -- SHA-256 based response validation
  • Dual environment -- Supports both test (cinv.ezpay.com.tw) and production (inv.ezpay.com.tw) environments

Prerequisites

  • Python 3.10 or higher
  • An ezPay merchant account with API credentials (MerchantID, HashKey, HashIV)
    • Test environment: register at https://cinv.ezpay.com.tw/
    • Production environment: register at https://inv.ezpay.com.tw/

Installation

Option 1: Install from PyPI

pip install mcp-ezpay-einvoice

Or with uv:

uv pip install mcp-ezpay-einvoice

Option 2: Run directly with uvx (no install needed)

EZPAY_MERCHANT_ID=your_id EZPAY_HASH_KEY=your_key EZPAY_HASH_IV=your_iv \
  uvx mcp-ezpay-einvoice

Option 3: Install from source

git clone https://github.com/asgard-ai-platform/mcp-ezpay-einvoice.git
cd mcp-ezpay-einvoice
uv venv && source .venv/bin/activate
uv pip install -e .

Configuration

Set the following environment variables before running the server:

Environment Variable Required Description
EZPAY_MERCHANT_ID Yes Your ezPay merchant ID
EZPAY_HASH_KEY Yes AES-256 encryption key (32 chars)
EZPAY_HASH_IV Yes AES-CBC initialization vector (16 chars)
EZPAY_IS_PRODUCTION No Set to true for production. Defaults to false (test environment)

You can also copy the .env.example file to .env and fill in your values:

cp .env.example .env

Usage

Claude Desktop

Add the following to your claude_desktop_config.json:

Using uvx (recommended -- no manual install needed):

{
  "mcpServers": {
    "ezpay-einvoice": {
      "command": "uvx",
      "args": ["mcp-ezpay-einvoice"],
      "env": {
        "EZPAY_MERCHANT_ID": "your_merchant_id",
        "EZPAY_HASH_KEY": "your_hash_key",
        "EZPAY_HASH_IV": "your_hash_iv",
        "EZPAY_IS_PRODUCTION": "false"
      }
    }
  }
}

Using pip install:

{
  "mcpServers": {
    "ezpay-einvoice": {
      "command": "mcp-ezpay-einvoice",
      "env": {
        "EZPAY_MERCHANT_ID": "your_merchant_id",
        "EZPAY_HASH_KEY": "your_hash_key",
        "EZPAY_HASH_IV": "your_hash_iv",
        "EZPAY_IS_PRODUCTION": "false"
      }
    }
  }
}

Claude Code

If you cloned the repo, the .mcp.json in the project root is auto-discovered by Claude Code. Just set the environment variables:

export EZPAY_MERCHANT_ID=your_merchant_id
export EZPAY_HASH_KEY=your_hash_key
export EZPAY_HASH_IV=your_hash_iv

Or add the server via CLI:

claude mcp add ezpay-einvoice -- uvx mcp-ezpay-einvoice

Cursor / Windsurf

Add to your MCP settings (.cursor/mcp.json or equivalent):

{
  "mcpServers": {
    "ezpay-einvoice": {
      "command": "uvx",
      "args": ["mcp-ezpay-einvoice"],
      "env": {
        "EZPAY_MERCHANT_ID": "your_merchant_id",
        "EZPAY_HASH_KEY": "your_hash_key",
        "EZPAY_HASH_IV": "your_hash_iv",
        "EZPAY_IS_PRODUCTION": "false"
      }
    }
  }
}

Running Directly

# With environment variables
EZPAY_MERCHANT_ID=your_id EZPAY_HASH_KEY=your_key EZPAY_HASH_IV=your_iv \
  python mcp_server.py

# Or with uvx
EZPAY_MERCHANT_ID=your_id EZPAY_HASH_KEY=your_key EZPAY_HASH_IV=your_iv \
  uvx mcp-ezpay-einvoice

Testing

# Validate credentials and connectivity
python scripts/auth/test_connection.py

# Run all 7 tool E2E tests
python tests/test_all_tools.py

Usage Examples

Once the MCP server is running, an AI assistant (e.g. Claude) can call these tools on your behalf. Here are real-world scenarios showing what you say and how the AI responds.


"Help me issue an invoice for a customer who bought a test product"

You: 幫我開一張發票,客人買了一個測試商品 105 元含稅,要印紙本發票

AI calls:

issue_invoice(
  merchant_order_no = "ORDER20260404001",
  buyer_name        = "Test Buyer",
  category          = "B2C",
  tax_type          = "1",
  tax_rate          = 5,
  amt               = 100,
  tax_amt           = 5,
  total_amt         = 105,
  item_name         = "Test Product",
  item_count        = "1",
  item_unit         = "pc",
  item_price        = "105",
  item_amt          = "105",
  print_flag        = "Y",
  status            = "1",
)

Result: SUCCESS -- Invoice TE00000000 issued, amount NT$105, random code 9393. The response includes BarCode, QRcodeL, and QRcodeR for printing.


"Issue a B2B invoice with the buyer's tax ID"

You: 幫我開一張 B2B 發票給統編 54352706 的公司,品名是顧問服務費 10000 元加稅 500 元

AI calls:

issue_invoice(
  merchant_order_no = "B2B20260404001",
  buyer_name        = "範例科技股份有限公司",
  buyer_ubn         = "54352706",
  category          = "B2B",
  tax_type          = "1",
  tax_rate          = 5,
  amt               = 10000,
  tax_amt           = 500,
  total_amt         = 10500,
  item_name         = "顧問服務費",
  item_count        = "1",
  item_unit         = "式",
  item_price        = "10000",
  item_amt          = "10000",
  print_flag        = "Y",
  status            = "1",
)

Result: SUCCESS -- B2B invoice issued with buyer UBN 54352706. The printed invoice includes BarCode and QR codes for tax reporting.


"Issue a B2C invoice with phone barcode carrier"

You: 客人要存手機條碼載具 /ABC1234,不印紙本,品項是拿鐵咖啡 2 杯 65 元、巧克力蛋糕 1 個 85 元

AI calls:

issue_invoice(
  merchant_order_no = "POS20260404001",
  buyer_name        = "顧客",
  category          = "B2C",
  carrier_type      = "0",
  carrier_num       = "/ABC1234",
  tax_type          = "1",
  tax_rate          = 5,
  amt               = 205,
  tax_amt           = 10,
  total_amt         = 215,
  item_name         = "拿鐵咖啡|巧克力蛋糕",
  item_count        = "2|1",
  item_unit         = "杯|個",
  item_price        = "65|85",
  item_amt          = "130|85",
  print_flag        = "N",
  status            = "1",
)

Result: SUCCESS -- Invoice stored to phone barcode carrier /ABC1234, no paper copy printed.


"Query the invoice I just issued"

You: 幫我查一下訂單 ORDER20260404001 的發票資料

AI calls:

query_invoice(
  search_type       = "1",
  merchant_order_no = "ORDER20260404001",
  total_amt         = 105,
)

Result: SUCCESS -- Returns full invoice details including invoice number TE00000000, buyer info, tax breakdown (Amt: 100, TaxAmt: 5, TotalAmt: 105), item details, invoice status (1=issued), upload status, BarCode, and QR codes.


"I need to issue a refund for a returned item"

You: 客人退了一杯拿鐵 65 元,幫我對發票 TE00000000 開折讓

AI calls:

issue_allowance(
  invoice_no        = "TE00000000",
  merchant_order_no = "ORDER20260404001",
  item_name         = "拿鐵咖啡",
  item_count        = "1",
  item_unit         = "杯",
  item_price        = "65",
  item_amt          = "65",
  item_tax_amt      = "3",
  total_amt         = 68,
  status            = "1",
)

Result: SUCCESS -- Allowance issued against invoice TE00000000 for NT$68 (65 + 3 tax). The remaining invoice amount is updated accordingly.


"Void that invoice -- it was issued by mistake"

You: 發票 TE00000000 開錯了,幫我作廢,原因是誤開發票

AI calls:

void_invoice(
  invoice_number = "TE00000000",
  invalid_reason = "誤開發票",
)

Result: SUCCESS -- Invoice TE00000000 voided. The platform will notify the buyer and upload the void record to the tax authority.

Note: Invoices can only be voided before the 14th of the next odd month (e.g., invoices issued in March-April can be voided before May 14th).


"Create a scheduled invoice for next week"

You: 幫我建一張預約發票,預計 2026-04-10 開立,金額 1000 元含稅

AI calls:

issue_invoice(
  merchant_order_no  = "SCH20260404001",
  buyer_name         = "預約客戶",
  category           = "B2C",
  tax_type           = "1",
  tax_rate           = 5,
  amt                = 952,
  tax_amt            = 48,
  total_amt          = 1000,
  item_name          = "訂閱服務",
  item_count         = "1",
  item_unit          = "式",
  item_price         = "1000",
  item_amt           = "1000",
  print_flag         = "Y",
  status             = "3",
  create_status_time = "2026-04-10",
)

Result: SUCCESS -- Invoice data saved, scheduled to issue on 2026-04-10. If you need to issue it earlier, use trigger_invoice to trigger it immediately.


"Trigger that scheduled invoice now"

You: 預約發票 SCH20260404001 要提前開立,幫我觸發

AI calls:

trigger_invoice(
  invoice_trans_no  = "26040423034604894",
  merchant_order_no = "SCH20260404001",
  total_amt         = 1000,
)

Result: SUCCESS -- Scheduled invoice triggered and issued immediately. Invoice number and random code are returned.

Tools Reference

Tool Description Key Parameters
issue_invoice Issue B2B/B2C e-invoice category, buyer_name, tax_type, amt, total_amt, item_name, item_count, item_price, print_flag
trigger_invoice Trigger deferred/scheduled invoice invoice_trans_no, merchant_order_no, total_amt
void_invoice Void an issued invoice invoice_number, invalid_reason
issue_allowance Issue allowance (credit note) invoice_no, item_name, item_count, item_price, total_amt
trigger_allowance Confirm or cancel pending allowance allowance_no, allowance_status (C=confirm, D=cancel)
void_allowance Void confirmed allowance allowance_no, invalid_reason
query_invoice Query invoice details search_type (0=by invoice number, 1=by order number)

Error Codes Reference

Common ezPay API response status codes:

Status Meaning
SUCCESS Operation completed successfully
LIB10001 Merchant ID does not exist
LIB10002 Encryption verification failed
LIB10003 Timestamp has expired
INV10001 Duplicate merchant order number
INV10003 Invoice number does not exist
INV20001 Invoice has already been voided
INV20002 Invoice void period has passed
ALW10001 Allowance amount exceeds invoice amount
ALW10002 Allowance number does not exist

For the complete error code list, refer to the ezPay API documentation.

Architecture

stdio (JSON-RPC 2.0)
  -> mcp_server.py           Entry point, side-effect import triggers tool registration
    -> app.py                FastMCP("mcp-ezpay-einvoice") singleton
      -> tools/invoice_tools.py    7 @mcp.tool() decorated functions
        -> connectors/ezpay_client.py   Encrypted form POST connector
          -> auth/ezpay_crypto.py       AES-256-CBC encrypt + CheckCode verify
            -> config/settings.py       Endpoints, URL builder, env var config
mcp-ezpay-einvoice/
|-- app.py                          # FastMCP server singleton
|-- mcp_server.py                   # Entry point (stdio transport)
|-- config/settings.py              # API endpoints, URL builder, credentials
|-- connectors/ezpay_client.py      # Encrypted form POST connector
|-- auth/ezpay_crypto.py            # AES-256-CBC encryption, CheckCode
|-- tools/invoice_tools.py          # 7 MCP tools (invoice + allowance lifecycle)
|-- tests/test_all_tools.py         # E2E test runner
|-- scripts/auth/test_connection.py # Connection and credential validator
|-- .mcp.json                       # Claude Code auto-discovery config
`-- .env.example                    # Environment variable template

Contributing

See CONTRIBUTING.md for setup instructions and guidelines.

  1. Fork the repository
  2. Create a feature branch: git checkout -b feat/your-feature
  3. Make changes and run tests
  4. Submit a pull request

License

MIT License -- see LICENSE for details.

Part of the Asgard Ecosystem

This server is part of the Asgard AI Platform, connecting AI assistants to real-world services across e-commerce, finance, government data, and more. See the full Asgard AI Platform.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured