Accounting Ledger MCP Server
# Accounting Ledger Model Context Protocol Server
A Model Context Protocol (MCP) server that provides accounting ledger creation, transaction and reporting capabilities using the [MCP Python SDK](https://github.com/modelcontextprotocol/python-sdk) and [Python Accounting](https://github.com/ekmungai/python-accounting) library.
## Overview
This MCP server enables AI assistants to interact with a double-entry accounting system, allowing for:
- Creating accounting entities including the chart of accounts, currencies, and tax codes
- Recording various types of transactions (cash sales, cash purchases, supplier invoices, customer bills)
- Generating financial reports (profit & loss statements)
## š¹ Demo Video
Watch the MCP Accounting Ledger Server in action:
<video src="https://github.com/user-attachments/assets/dae6e6ad-03cc-40ff-ae30-45d441bd89b2"></video>
*This demo shows the complete workflow from setup to creating ledgers, recording transactions, and generating reports using the MCP server with Claude Desktop.*
## Features
### š¢ Entity Management
- Create accounting entities with default configuration (companies/organizations)
### š° Transaction Recording
- **Cash Sales**: Record immediate revenue transactions
- **Cash Purchases**: Record immediate expense transactions
- **Client Invoices**: Create receivable transactions
- **Supplier Bills**: Create payable transactions
### š Financial Reporting
- Generate profit & loss statements
- Customizable date ranges for reports
### š§ Built-in Tools
- Echo tool for testing connectivity
- Comprehensive error handling
- SQLite database backend
## Installation
This project uses [UV](https://docs.astral.sh/uv/) for dependency management. Make sure you have UV installed on your system.
### Prerequisites
Install uv
### Setup
1. **Clone the repository**:
Clone this repository locally and update config.toml with the path to your local project
2. **Create the virtual environment**:
```bash
uv venv
```
3. **Activate the virtual environment**:
```bash
source .venv/bin/activate
```
4. **Install dependencies**:
```bash
uv sync
```
## Running the Server
### Development Mode
To run the server in development mode:
```bash
uv run mcp dev server.py
```
This will start the server the MCP Inspector which is a useful tool for testing and debugging the MCP server. To access the MCP Inspector navigate to `http://localhost:6274`.
### Testing in Claude Desktop
After installing Claude Desktop add the MCP server to it by running:
```bash
uv run mcp install server.py
```
After running this command open Claude Desktop. There will most likely be an error with the MCP server. Open Claude Desktop settings, navigate to Developer, select our MCP Server (My Ledger) and click Edit Config. Open the JSON configuration file and update the command and args values as shown (replacing path_to_server with the path to the server on your local machine).
```json
{
"mcpServers": {
"My Ledger": {
"command": "/path_to_server/.venv/bin/python",
"args": [
"/path_to_server/server.py"
]
}
}
}
```
Now restart Claude Desktop and the MCP server should be working.
## Available Tools
| Tool | Description | Parameters |
|------|-------------|------------|
| `echo` | Test connectivity | `message: str` |
| `create_ledger` | Set up new accounting entity | `entity_name: str` |
| `record_cash_sale` | Record cash sale transaction | `narration: str, amount: float, quantity: int, tax_code: str, entity_name: str` |
| `record_cash_purchase` | Record cash purchase transaction | `narration: str, amount: float, quantity: int, tax_code: str, entity_name: str` |
| `record_client_invoice` | Create client invoice | `narration: str, amount: float, quantity: int, tax_code: str, entity_name: str` |
| `record_supplier_bill` | Create supplier bill | `narration: str, amount: float, quantity: int, tax_code: str, entity_name: str` |
| `generate_profit_loss_report` | Generate P&L report | `entity_name: str, start_date: str, end_date: str` |
## Configuration
The server uses a `config.toml` file for configuration. Key settings include:
- **Database**: SQLite database URL
- **Account Types**: Chart of accounts configuration
- **Transaction Types**: Available transaction types and prefixes
- **Tax Settings**: Default tax codes and rates
- **Reporting**: Report formatting and sections
## Database Schema
The system uses the Python Accounting library's database schema, which includes:
- **Entities**: Companies/organizations
- **Currencies**: Supported currencies
- **Accounts**: Chart of accounts with types
- **Taxes**: Tax codes and rates
- **Transactions**: Various transaction types
- **Line Items**: Transaction details
## Development
### Project Structure
```
mcpServerDemo/
āāā server.py # Main MCP server implementation
āāā main.py # Entry point
āāā pyproject.toml # UV project configuration
āāā config.toml # Accounting system configuration
āāā test_accounting.db # SQLite database
āāā python_accounting/ # Local Python Accounting library
```
### Adding New Features
1. Add new tools to `server.py` using the `@mcp.tool()` decorator
2. Update the configuration in `config.toml` if needed
3. Test using the MCP Inspector
4. Update this README with new tool documentation
## Dependencies
- **mcp[cli]**: Model Context Protocol SDK
- **python-dateutil**: Date parsing utilities
- **sqlalchemy**: Database ORM
- **strenum**: String enumerations
- **toml**: Configuration file parsing
TDQS
Scored across 7 tools
Each tool has a clearly distinct purpose: setup, reporting, and four specific transaction types. The echo tool is unrelated but not confusing, as it serves a generic utility function.
Most tools follow a consistent verb_noun pattern with snake_case (create_ledger, generate_profit_loss_report, record_*). The exception is 'echo', which is a simple verb, but this is a minor deviation.
Seven tools is well-scoped for an accounting ledger server: one setup tool, four transaction recording tools, one reporting tool, and one utility. Each tool earns its place.
The server covers setup, transaction recording, and reporting, but lacks fundamental operations like listing or retrieving transactions, editing, or deleting records. This creates a gap where agents cannot inspect or manage recorded data directly.