Expense Tracker MCP Server
Enables managing personal expenses, budgets, and financial summaries through natural language, with tools for adding, listing, updating, deleting expenses, setting budgets and credits, and generating dashboards.
README
š° Multi-User Expense Tracker MCP Server
A production-ready Async Multi-User Expense Tracker MCP Server built using FastMCP, PostgreSQL (Neon), and Python.
This server is designed to support multiple users, isolating budgets, expenses, and monthly finance per user via their email address (auto-creating users dynamically on first use). It validates categories and subcategories dynamically using a localized configuration file, and isolates credit card self-usage from borrowed/lent expenses.
š Key Features
š¤ 1. Zero-Config Multi-Tenant Isolation
- Dynamic Onboarding: No complex sign-up flows, passwords, or authentication overhead. The first time a user's email address is supplied in a tool call, a new user account is generated automatically.
- Strict Data Isolation: Every database query is restricted using
WHERE user_id = %s. It is impossible for a user to view, edit, or delete another user's financial details. - Race-Condition Free: Uses atomic
INSERT ... ON CONFLICT DO NOTHINGSQL followed by a SELECT query inside transactions to prevent duplicate user creation under high concurrent loads.
š·ļø 2. Dynamic Category & Subcategory Validation
- Local Rules Engine: Uses [category.json](file:///Users/sarthakshukla/Desktop/MCP/category.json) as the source of truth for allowed expense classifications.
- Smart Normalization: Automatically normalizes input casing, strips trailing/leading whitespaces, and converts spaces to underscores (e.g., inputting
"Food"and"Dining Out"normalizes cleanly to"food"and"dining_out"). - Tool Discovery: Exposes an MCP resource
expense://categoriescontaining the JSON config. This lets AI clients query and inspect valid classifications before calling tools.
š³ 3. Borrowed Expense Isolation (Credit Card & Cash)
- Separate Summaries: Expenses flagged as borrowed are excluded from the user's monthly budget, personal spend, and average transaction stats.
- Auto-Detection Rules: The system automatically detects a borrowed expense if:
- The category is
"credit_card_usage"and the subcategory is"borrowed_to_friend". - The note contains keywords like
"borrowed to friend","borrowed to a friend", or"lent to"(case-insensitive).
- The category is
- Virtual Categorization: During monthly summary generations, all borrowed expenses are grouped and returned under a separate virtual category called
"borrowed", allowing clean data visualization.
š ļø Tech Stack
- Python 3.11+
- FastMCP: Modern tool-first MCP server framework.
- PostgreSQL (Neon-compatible): High-performance relational database.
- psycopg3: Modern, type-safe async driver for PostgreSQL.
- psycopg-pool: Connection pooler designed for high concurrent throughput.
- Pydantic: Robust runtime model typing and constraint validation.
- uv: Next-generation Python package management.
š Project Structure
expense-tracker/
ā
āāā app/
ā āāā config.py # Loads database settings & pool sizing from environment
ā āāā databases.py # Configures & lifecycle-manages psycopg connection pools
ā āāā schema.py # Defines table queries, keys, indexes, and cascades
ā āāā models.py # Contains Pydantic models for validation and type-hints
ā āāā resources.py # Implements category loading and validation/normalization
ā ā
ā āāā repository/ # Database query operations (zero business logic)
ā ā expense_repository.py
ā ā finance_repository.py
ā ā user_repository.py
ā ā
ā āāā services/ # Orchestrates business logic, validations, & orchestrates transactions
ā expense_service.py
ā finance_service.py
ā summary_service.py
ā
āāā category.json # Source of truth for categories & subcategories
āāā main.py # Instantiates FastMCP, sets tool endpoints, & binds lifespan
āāā pyproject.toml # Declares project dependencies & tool mappings
āāā uv.lock # Deterministic dependency resolution lockfile
āāā README.md # Project documentation
š Database Schema Details
1. users
Represents the primary tenant identification.
CREATE TABLE users (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
email TEXT UNIQUE NOT NULL,
full_name TEXT DEFAULT '',
created_at TIMESTAMPTZ DEFAULT NOW()
);
2. expenses
Stores transaction items linked to users.
CREATE TABLE expenses (
id SERIAL PRIMARY KEY,
user_id UUID NOT NULL REFERENCES users(id) ON DELETE CASCADE,
date DATE NOT NULL,
amount NUMERIC(12,2) NOT NULL,
category TEXT NOT NULL,
subcategory TEXT DEFAULT '',
note TEXT DEFAULT '',
is_borrowed BOOLEAN DEFAULT FALSE
);
-- Optimize queries by user, dates, categories, and borrow status
CREATE INDEX idx_expense_user ON expenses(user_id);
CREATE INDEX idx_expense_user_date ON expenses(user_id, date);
CREATE INDEX idx_expense_user_category ON expenses(user_id, category);
CREATE INDEX idx_expense_user_borrowed ON expenses(user_id, is_borrowed);
3. monthly_finance
Stores budget limits and credit allocations.
CREATE TABLE monthly_finance (
user_id UUID NOT NULL REFERENCES users(id) ON DELETE CASCADE,
month DATE NOT NULL,
budget NUMERIC(12,2) DEFAULT 0,
credit NUMERIC(12,2) DEFAULT 0,
PRIMARY KEY (user_id, month)
);
CREATE INDEX idx_finance_user_month ON monthly_finance(user_id, month);
āļø Setup & Installation
1. Re-initialize Database (Breaking Schema Change)
If you had an old version of the database running, you must drop the old tables first:
DROP TABLE IF EXISTS expenses, monthly_finance, users CASCADE;
2. Setup Environment Variables
Create a .env file in the root directory:
DATABASE_URL=postgresql://neondb_owner:********@ep-example.us-east-1.aws.neon.tech/neondb?sslmode=require
POOL_MIN_SIZE=2
POOL_MAX_SIZE=10
3. Synchronize Dependencies
Install python packages using uv:
uv sync
š Running the Server
Start the server using uv run:
uv run fastmcp run main.py:mcp --transport http --host 0.0.0.0 --port 8000
This launches a SSE/HTTP server listening at http://localhost:8000/mcp.
š ļø MCP Tools Reference
Every tool automatically takes email: str as the first parameter to identify the caller and scope operations.
1. add_expense
Adds a new expense.
- Arguments:
email(str): Email address of the user.date(str): Date formatted as"YYYY-MM-DD".amount(float): Positive numeric value.category(str): Key insidecategory.json.subcategory(str): Sub-item matching category.note(str): Optional description.is_borrowed(bool, optional): Force flags the transaction as borrowed.
2. list_expenses
Fetch transactions for a given date range.
- Arguments:
email(str)start_date(date)end_date(date)
3. update_expense
Update a transaction (scopes updates to user owning the expense_id).
- Arguments:
email(str)expense_id(int)date(str)amount(float)category(str)subcategory(str)note(str)is_borrowed(bool)
4. delete_expense
Permanently delete an expense.
- Arguments:
email(str)expense_id(int)
5. recent_expenses
Get latest transactions.
- Arguments:
email(str)limit(int, default=10)
6. set_monthly_budget / set_monthly_credit
Update financial targets for a specific month (composite PK handles conflicts).
- Arguments:
email(str)month(str, format"YYYY-MM")budget/credit(float)
7. summarize
Generates a comprehensive financial overview of the month.
-
Arguments:
email(str)start_date(date)end_date(date)category(str, optional)
-
Example Output:
{ "status": "ok", "month": "2026-07", "budget": 50000.0, "credit": 70000.0, "total_expense": 23500.0, // Excludes borrowed amounts "total_borrowed": 1500.0, // Tracks borrowed separately "remaining_budget": 26500.0, "remaining_credit": 46500.0, "budget_used_percent": 47.0, "is_over_budget": false, "over_budget_amount": 0.0, "transactions": 38, "average_transaction": 618.42, "average_daily_spend": 758.06, "largest_category": "food", "largest_category_amount": 12000.0, "category_summary": [ { "category": "food", "total_amount": 12000.0 }, { "category": "transport", "total_amount": 8000.0 }, { "category": "borrowed", "total_amount": 1500.0 } ] }
š¤ Integrating with Clients
Claude Desktop Configuration
Add the following block to your claude_desktop_config.json:
{
"mcpServers": {
"expense-tracker": {
"command": "uv",
"args": [
"--directory",
"/absolute/path/to/expense-tracker",
"run",
"fastmcp",
"run",
"main.py:mcp"
]
}
}
}
Note: Make sure to replace /absolute/path/to/expense-tracker with the actual path to your directory.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.