Expense Tracker MCP Server
Tracks personal expenses and budgets via MCP, storing entries in SQLite and supporting category-based queries.
README
Expense Tracker MCP Server
This project is a lightweight FastMCP server for managing personal expenses and budgets over HTTP. It provides a simple way to store expenses in a local SQLite database, organize them by category, and query spending information through MCP-compatible tools.
What it does
- Tracks expense entries with amounts, dates, and descriptions
- Supports category-based organization using a JSON category list
- Exposes expense-related operations through an MCP server endpoint
- Can be run locally for development or deployed remotely for HTTP access
Users can view their data through MCP clients such as Claude Desktop:
- Ask "show all expenses" to call
list_expenses. - Ask "show my budgets" to call
list_budgets. - Ask for a date range, such as "show expenses from 2026-08-01 to 2026-08-31", to filter the expense list.
Run locally
uv run main.py
The local MCP endpoint is:
http://localhost:8000/mcp
Remote deployment
This project is already configured for a remote HTTP MCP server:
mcp.run(transport="http", host="0.0.0.0", port=port)
The server reads these environment variables:
| Variable | Default | Purpose |
|---|---|---|
HOST |
0.0.0.0 |
Network interface to bind |
PORT |
8000 |
HTTP port, usually set by the hosting provider |
EXPENSE_DB_PATH |
./expenses.db |
SQLite database path |
EXPENSE_CATEGORIES_PATH |
./categories.json |
Categories JSON path |
For FastMCP Cloud, set EXPENSE_DB_PATH to a writable persistent path if your
deployment provides one. If you do not set it and the app directory is
read-only, the server falls back to a temp directory so writes can succeed, but
that data may disappear after the cloud server restarts.
Typical remote start command:
uv run main.py
After deployment, connect your MCP client to:
https://your-domain.example/mcp
Important remote notes
- Use HTTPS for public access.
- Make sure the hosting provider exposes the assigned
PORT. - If you keep SQLite, configure persistent storage and set
EXPENSE_DB_PATHto that mounted path. Otherwise the database may reset after redeploys. - Use the
database_statustool from Claude to see the active database path and whether SQLite can be initialized. - For a public server, add authentication or place it behind a protected gateway before storing real financial data.
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。