hledger-mcp
MCP server for hledger that provides tools for financial reports, transaction management, and account management.
README
hledger MCP Server
A Model Context Protocol (MCP) server that provides tools for interacting with the hledger plain text accounting system.
Features
This MCP server provides the following hledger tools:
Financial Reports
- hledger_balance - Generate balance reports with filtering options
- hledger_incomestatement - Generate income statements
- hledger_balancesheet - Generate balance sheets
- hledger_cashflow - Generate cash flow statements
Transaction Management
- hledger_register - View transaction listings with filtering
Account Management
- hledger_accounts - List all accounts with optional filtering
- hledger_stats - Show journal statistics
Installation
- Make sure you have hledger installed on your system
- Clone this repository
- Install dependencies:
bun install
Configuration for VS Code
When using this MCP server with VS Code, you need to ensure it runs hledger commands in the correct directory. Add the server to your VS Code MCP settings (usually in your VS Code settings):
{
"mcpServers": {
"hledger": {
"command": "bun",
"args": ["run", "/absolute/path/to/hledger-mcp/index.ts"],
"env": {
"LEDGER_FILE": "/path/to/your/finances/main.journal"
}
}
}
}
File Resolution Priority
The server resolves which hledger file to use in this order:
- Explicit
fileparameter - if provided in the tool call LEDGER_FILEenvironment variable - if set- hledger's default file discovery - follows hledger's standard rules
Working Directory Handling
The MCP server supports several ways to specify which hledger files to use:
Option 1: Use LEDGER_FILE Environment Variable (Recommended)
Set the LEDGER_FILE environment variable in your MCP server configuration. The server will automatically use the journal file specified in $LEDGER_FILE.
Option 2: Use the workingDirectory Parameter
All tools accept an optional workingDirectory parameter that specifies where hledger commands should be executed.
Option 3: Use Absolute File Paths
You can specify the exact path to your journal file using the file parameter.
Option 4: Use Relative Paths with Working Directory
If your journal files are in a specific directory structure, you can combine the workingDirectory parameter with a relative file path.
Usage
As an MCP Server
Run the server in stdio mode:
bun run index.ts
Tool Parameters
Common Parameters
- file (optional): Path to specific hledger journal file (overrides
$LEDGER_FILEenv var) - workingDirectory (optional): Directory to run hledger commands from (defaults to current directory)
- period (optional): Time period filter using natural language (e.g., "this month", "2024", "last quarter")
- beginDate (optional): Precise begin date for filtering (YYYY-MM-DD, YYYY-MM, or YYYY)
- endDate (optional): Precise end date for filtering (YYYY-MM-DD, YYYY-MM, or YYYY)
- depth (optional): Account depth limit for reports
Note: Use beginDate/endDate for precise date ranges and monthly averages. If both period and beginDate/endDate are provided, the specific dates take precedence.
Environment Variables
LEDGER_FILE: Default journal file path (used when no explicit file is specified)
hledger_balance
- accounts (optional): Account pattern to filter
- period (optional): Natural language time period (e.g., "this month", "2024")
- beginDate (optional): Begin date for filtering (YYYY-MM-DD or YYYY-MM or YYYY)
- endDate (optional): End date for filtering (YYYY-MM-DD or YYYY-MM or YYYY)
- monthly (optional): Show monthly breakdown
- average (optional): Show averages over the reporting period
- flat (optional): Show flat account list instead of tree
hledger_register
- query (optional): Query to filter transactions
- period (optional): Natural language time period (e.g., "this month", "2024")
- beginDate (optional): Begin date for filtering (YYYY-MM-DD or YYYY-MM or YYYY)
- endDate (optional): End date for filtering (YYYY-MM-DD or YYYY-MM or YYYY)
- monthly (optional): Show monthly breakdown
- width (optional): Output width (default: 80)
hledger_accounts
- pattern (optional): Pattern to filter accounts
- tree (optional): Show accounts in tree format
Requirements
Development
The server is built using:
- Bun for JavaScript runtime and process execution
- Model Context Protocol SDK for MCP server functionality
- Zod for input validation
To run in development mode:
bun run index.ts
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。