---
title: "Getting the most out of BankBridge's 12 MCP tools"
slug: getting-the-most-out-of-12-mcp-tools
url: https://bankbridge.money/guides/getting-the-most-out-of-12-mcp-tools
category: tip
published: 2026-04-23
updated: 2026-04-23
---
# Getting the most out of BankBridge's 12 MCP tools

> BankBridge exposes 12 MCP tools covering accounts, transactions, spending aggregates, recurring charges, cashflow, merchant history, investments, and bank connection. Your agent picks the right one automatically most of the time — but knowing what each tool does helps you phrase questions that get cleaner answers.

Getting the Most From 12 MCP Tools | BankBridge | BankBridge 

[Tip](/guides/tips)

# Getting the most out of BankBridge's 12 MCP tools

Updated Apr 23, 2026·8 min read

BankBridge exposes 12 MCP tools covering accounts, transactions, spending aggregates, recurring charges, cashflow, merchant history, investments, and bank connection. Your agent picks the right one automatically most of the time — but knowing what each tool does helps you phrase questions that get cleaner answers.

## Depository + credit

Nine tools for checking, savings, and credit-card data. This is where most people spend most of their time.

### list\_accounts

Balances and types for every connected account.

> “What do I have in each of my accounts right now?”

### get\_account

Detail lookup for a single account by ID. Usually invoked as a follow-up after list\_accounts.

> “Tell me more about the checking account — official name, mask, type.”

### list\_transactions

The workhorse. Filter by date, amount, category, account; paginate. This tool powers most questions.

> “Show me every charge between $100 and $500 this month.”

### search\_transactions

Substring match on merchant name. Use when you know part of the name but the category might be wrong.

> “Find every transaction with 'coffee' in the merchant name.”

### get\_spending\_summary

Group spend by category, merchant, month, or week. Returns sorted buckets.

> “Break down my March spending by merchant, top 10.”

### get\_recurring\_charges

Pattern-detected subscriptions + utilities. Similar merchant + similar amount at similar cadence.

> “What's the full list of my recurring charges?”

### get\_monthly\_cashflow

Income vs expenses for a specific month plus top sources + categories. One call, one month.

> “Cashflow for March 2026. Include top expense categories.”

### get\_merchant\_history

Every charge for one merchant. Great for 'how much have I spent at X' questions.

> “All charges at Whole Foods this year, with the monthly total.”

### list\_categories

Unique categories present in your data. Useful for discovery ('what categories are even available').

> “Which spending categories show up in my data?”

## Investment

Two tools for brokerage data — holdings (snapshot) and transactions (history).

### list\_holdings

Current positions with gain/loss against cost basis.

> “Every position I hold, sorted by current value, with gain/loss.”

### list\_investment\_transactions

Buys, sells, dividends, fees. Filter by date, account, type.

> “All dividend payments in the last 90 days.”

## Connection

One tool for when the agent decides a new bank connection is needed.

### connect\_bank

Returns a URL that opens the bank-link flow. Use when list\_accounts returns no connections or the user asks to add a bank.

> “I want to connect another bank — what do I do?”

## FAQ

Do I need to name tools when asking the agent?

No, and you shouldn't. The agent is trained to pick the right tool based on your natural-language question. Naming the tool explicitly is useful for debugging (when the agent picks wrong) or teaching (when you're showing someone how it works).

Why don't you have a single do-everything tool?

Specific tools get specific answers faster and with less context. An agent calling three focused tools in sequence is more efficient than one tool that needs to guess what you want.

## FAQ

### Do I need to name tools when asking the agent?

No, and you shouldn't. The agent is trained to pick the right tool based on your natural-language question. Naming the tool explicitly is useful for debugging (when the agent picks wrong) or teaching (when you're showing someone how it works).

### Why don't you have a single do-everything tool?

Specific tools get specific answers faster and with less context. An agent calling three focused tools in sequence is more efficient than one tool that needs to guess what you want.