Accounting

This MCP server provides a personal accounting tool that exposes finance management capabilities to an AI assistant. It lets you record transactions, check balances, review transaction histories, and generate monthly summaries, all through a small, JSON-RPC powered interface. You can interact with it using an MCP client to manage your finances via natural language or scripted requests, making it easy to build budgeting, expense tracking, and financial insights into your AI workflows.

How to use

You can use an MCP client to connect to the accounting MCP server and perform core actions: add a new transaction, query the current balance, list or filter transactions, and fetch the monthly financial summary. Treat these actions as standard tools available to your AI assistant. You’ll send requests to the server through the MCP interface, and the server will respond with confirmations, balances, or lists that you can display to the user.

How to install

Prerequisites: you need Python 3.11 or newer and a tool to manage environments, such as conda or a virtual environment. You should also have a method to install Python dependencies (pip is assumed here).

Step 1: Set up your environment. You can use conda for a quickStart or a plain virtual environment. Choose one of the following options.

# Option A: using conda (recommended for speed)
conda create -n accounting-mcp python=3.11
conda activate accounting-mcp

# Install runtime dependencies with uv (fast MCP tooling)
uv pip install -r requirements.txt
uv pip install -r requirements-dev.txt

# Option B: pure conda (no uv)
conda create -n accounting-mcp python=3.11
conda activate accounting-mcp

pip install -r requirements.txt
pip install -r requirements-dev.txt

# Option C: pure virtual environment
python -m venv venv
source venv/bin/activate  # macOS/Linux

pip install -r requirements.txt
pip install -r requirements-dev.txt

Step 2: Run the MCP server locally. Start the accounting MCP server using Python and the server module provided with the project.

python -m accounting_mcp.server

Optional: You can also run an interactive test client to try commands against the server while developing or validating behavior.

python test_client.py

Additional notes

If you want to simulate the exact run using a ready-made configuration, you can leverage a preconfigured MCP setup that invokes the server module directly via a Python command. This approach shows how the server can be started in an embedded development scenario where the MCP client and tools are orchestrated from the same environment.

Available tools

add_transaction

Adds a new income or expense record with details such as amount, category, and description.

get_balance

Retrieves the current account balance based on recorded transactions.

list_transactions

Lists transactions with optional filters such as date range or category.

get_monthly_summary

Provides a monthly financial summary including totals by category and net balance.