mcp-server-ledger: A Ledger CLI MCP Server
Overview
A Model Context Protocol server for interacting with Ledger CLI, a powerful, double-entry accounting system accessible from the command line. This server allows Large Language Models to query and analyze your financial data through the Ledger CLI tool.
This MCP server exposes Ledger CLI's functionality through a standardized interface, making it easy for AI assistants to help you with financial reporting, budget analysis, and accounting tasks.
Features
The server provides the following tools that map to Ledger CLI commands:
ledger_balance
* Shows account balances with powerful filtering options
* Inputs: query pattern, date ranges, display options
* Returns: Formatted balance report
ledger_register
* Shows transaction register with detailed history
* Inputs: query pattern, date ranges, sorting options
* Returns: Formatted register report
ledger_accounts
* Lists all accounts in the ledger file
* Input: optional query pattern
* Returns: List of matching accounts
ledger_payees
* Lists all payees from transactions
* Input: optional query pattern
* Returns: List of matching payees
ledger_commodities
* Lists all commodities (currencies) used
* Input: optional query pattern
* Returns: List of matching commodities
ledger_print
* Prints transactions in ledger format
* Inputs: query pattern, date ranges
* Returns: Formatted ledger entries
ledger_stats
* Shows statistics about the ledger file
* Input: optional query pattern
* Returns: Statistical summary of the ledger
ledger_budget
* Shows budget analysis
* Inputs: query pattern, date ranges, reporting period
* Returns: Budget report
ledger_raw_command
* Runs a raw Ledger CLI command
* Input: command arguments as a list of strings
* Returns: Command output as text
Prerequisites
- Ledger CLI must be installed and available in your PATH
- A valid Ledger file with your financial data
Installation
Using uv (recommended)
The easiest way to install and run this server is with uv:
uv sync
Configuration
The server requires a path to your Ledger file. This can be provided through:
- The
LEDGER_FILE environment variable
- Command-line arguments when starting the server
Using with Claude Desktop
Add this to your claude_desktop_config.json:
"mcpServers": {
"ledger": {
"command": "uv",
"args": [
"run",
"--env-file",
"/path/to/your/.env", // Optional: to set LEDGER_FILE
"--with",
"mcp[cli]",
"mcp",
"run",
"<repo_path>/main.py" // Path to the main.py file in this repository
]
}
}
Usage Examples
Once configured, you can ask your AI assistant questions about your financial data:
- "Show me my expenses for the last month"
- "What's my current balance in all accounts?"
- "List all transactions with Amazon"
- "How much did I spend on groceries in 2023?"
- "Show me my budget performance for Q1"
The AI will use the appropriate Ledger CLI commands through the server to get the information.
Debugging
For more detailed local testing:
mcp dev main.py
Development
This server is built using the Model Context Protocol Python SDK. For more information about MCP and how to develop MCP servers, see the official documentation.
To contribute to this project:
- Clone the repository
- Install development dependencies
- Make your changes
- Test using the MCP inspector or by integrating with Claude Desktop
Security Considerations
This server runs Ledger CLI commands on your financial data. While it includes basic validation to prevent command injection, you should:
- Only use with trusted clients
- Be careful about which file paths you expose
- Review all commands before execution
License
This MCP server is licensed under the MIT License. Feel free to use, modify, and distribute it according to the license terms.
