local-expense-tracker
A local MCP server for tracking personal expenses using SQLite, enabling users to add, list, and summarize expenses via natural language.
README
Expense Tracker -- Local MCP Server
A simple local MCP (Model Context Protocol) server for tracking personal expenses using SQLite. This server exposes expense-related tools (add, list, summarize) that can be consumed by MCP-compatible clients such as Claude Desktop, Cursor, or other LLM hosts.
No HTTP server. No frontend. Just a clean local tool backend for LLMs.
Features
- Add expense entries with date, amount, category, subcategory, and notes
- List expenses within a date range
- Summarize expenses by category
- Local SQLite database (auto-created)
- Categories exposed as an MCP resource (
expense://categories) - Zero configuration beyond Python
Project Structure
expense_tracker/
│
├── server.py # MCP server code
├── categories.json # Expense categories (editable at runtime)
└── expenses.db # SQLite database (auto-created)
Requirements
-
Python 3.9+
-
uv(recommended) -
fastmcp
Setup
1. Install uv (once)
pip install uv
Verify installation:
uv --version
2. Create a virtual environment
From the project directory:
uv venv
Activate it:
-
Windows:
.venv\Scripts\activate -
macOS / Linux:
source .venv/bin/activate
3. Install dependencies
uv pip install fastmcp
Running the MCP Server
Start the server using:
uv run python server.py
Note: You will see no output. This is expected behavior as the server runs silently and waits for a compatible client (like Claude) to connect.
To stop the server: CTRL + C
Database Behavior
-
The
expenses.dbfile is automatically created on first run. -
Tables are created using
CREATE TABLE IF NOT EXISTS. -
No manual database setup is required.
Available MCP Tools
add_expense
Adds a new expense entry to the database.
-
date: string (format:
YYYY-MM-DD) -
amount: number
-
category: string
-
subcategory: string (optional)
-
note: string (optional)
list_expenses
Lists all expenses within a given date range.
-
start_date: string (
YYYY-MM-DD) -
end_date: string (
YYYY-MM-DD)
summarize
Summarizes expenses by category.
-
start_date: string (
YYYY-MM-DD) -
end_date: string (
YYYY-MM-DD) -
category: string (optional)
MCP Resources
expense://categories
Returns the contents of categories.json. The file is read fresh on every request, allowing category updates without restarting the server.
Using with Claude Desktop
Add the server to your claude_desktop_config.json (found in %APPDATA%\Claude on Windows or ~/Library/Application Support/Claude on macOS):
{
"mcpServers": {
"expense-tracker": {
"command": "uv",
"args": [
"run",
"--path",
"C:/absolute/path/to/expense_tracker",
"python",
"server.py"
]
}
}
}
Note: Ensure you use absolute paths and forward slashes / even on Windows.
License
MIT License --- free to use, modify, and distribute.
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。