Expense Tracker MCP Server

An MCP server for tracking expenses with a small SQLite-backed data store and a built-in categories resource.

What it provides

• add_expense: insert a new expense record.

• list_expenses: list expenses in an inclusive date range.

• summarize: aggregate spending by category for a date range, with an optional category filter.

• expense:///categories: expose the category list as JSON.

Related MCP server: local-expense-tracker

Storage

The server stores data in a SQLite database inside the operating system temporary directory:

• Database file: expenses.db

• Location: the current temp folder returned by Python's tempfile.gettempdir()

On startup, the server creates the expenses table if it does not already exist and performs a small write test to confirm the database is writable.

Categories

If src/expense_tracker_mcp_server/categories.json exists, the server serves that file content through expense:///categories.

If the file is missing, it falls back to these default categories:

• Food & Dining

• Transportation

• Shopping

• Entertainment

• Bills & Utilities

• Healthcare

• Travel

• Education

• Business

• Other

Requirements

• Python 3.14 or later

• fastmcp

• aiosqlite

Run the server

From the project root:

python src/expense_tracker_mcp_server/main.py

The server starts an HTTP transport on 0.0.0.0:8000.

Tool reference

add_expense

Adds a new expense entry.

Arguments:

• date: expense date as text

• amount: numeric amount

• category: top-level category name

• subcategory: optional subcategory text

• note: optional note text

Returns:

• status

• id on success

• message

list_expenses

Lists expenses whose date value falls between start_date and end_date, inclusive.

Arguments:

• start_date

• end_date

Returns a list of expense objects ordered by newest date first, then by descending id.

summarize

Summarizes expenses by category for a date range.

Arguments:

• start_date

• end_date

• category optional filter

Returns totals grouped by category, including sum of amount and count.

Notes

• The server currently uses a temporary SQLite database, so data is not meant to be permanent.

• The module is implemented in src/expense_tracker_mcp_server/main.py.