Skip to main content

​
Overview

Tight Bank Transactions MCP tools enable AI assistants and agents to search, review, and update bank transactions. This allows users to categorize transactions, add notes, and request bookkeeper reviews, all through conversational AI interactions. With these tools, you can build AI-powered accounting experiences where users ask questions like “Show me all uncategorized transactions from last month” or “Categorize this coffee shop expense as a recruitment expense.”

​
Available Tools

​
Query Bank Transactions

The query_bank_transactions tool searches for bank transactions based on multiple criteria. You can filter transactions by amount, date, description, vendor, category, review status, and more.
​
amount
number | range
Search for transactions with a specific amount or within an amount range (e.g., “$50-$100”)
​
date
date | date range
Filter transactions by specific date or date range (e.g., “last month”, “January 2025”, “2025-01-01 to 2025-01-31”)
​
description
string
Fuzzy search on the bank description field
​
notes
string
Fuzzy search on user-added notes attached to transactions
​
classSegment
string
Filter transactions belonging to specific class segments
​
vendor
string
Search for transactions associated with specific vendors or customers
​
category
string
Filter transactions by accounting category or account (e.g., “Meals & Entertainment”, “Office Supplies”)
​
splits
object
Search for transactions that are split across multiple categories or amounts
​
reviewStatus
string
Filter by review status (e.g., “needs review”, “reviewed by bookkeeper”, “pending bookkeeper review”)
​
type
string
Filter by transaction type: Expense, Revenue, Transfer, Tax Payment, Invoice Payment, Bill Payment, etc.
You can combine multiple search criteria to narrow results. For example, search for “uncategorized expenses over $100 from last week” by combining amount, date, category, and type filters.

​
Get Bank Transaction Details

The get_bank_transaction tool retrieves detailed information about a specific bank transaction. Use this tool to inspect transaction details before making updates or to discuss categorization decisions with the user. This tool returns comprehensive transaction data including:
  • Transaction amount and date
  • Bank description and user-added notes
  • Current categorization (vendor, category, classSegment)
  • Split information if applicable
  • Review status and history
  • Related invoice or bill payments
  • Attachments and receipts
This tool is essential for conversational workflows where users want to discuss a specific transaction with the AI before making changes. The detailed context helps the AI provide informed categorization suggestions.

​
Update Bank Transaction

The update_bank_transaction tool modifies any field on a bank transaction. This is the primary tool for categorizing expenses, adding notes, splitting transactions, and managing review workflows.
​
classSegment
string
Assign or update the business class segment for tracking transactions across different parts of the business
​
vendor
string
Set or update the vendor or customer associated with the transaction
​
category
string
Update the accounting category or account for the transaction (e.g., “Travel”, “Office Supplies”, “Professional Services”)
​
splits
array
Split a transaction across multiple categories, classSegments, or customers. Each split can have its own amount, category, and allocation.
​
reviewStatus
string
Request review from a bookkeeper or mark a transaction as reviewed
​
notes
string
Add or update notes on the transaction for additional context or documentation

​
Common Workflows

​
Categorizing Transactions Needing Review

One of the most common workflows is identifying transactions that need categorization and applying the appropriate category:
  1. Query for transactions needing review using query_bank_transactions with the reviewStatus filter set to “NEEDS_REVIEW”
  2. Review each transaction using get_bank_transaction to understand the context
  3. Update the transaction with the appropriate category using update_bank_transaction
This workflow helps business owners stay on top of their bookkeeping by ensuring all transactions are properly categorized for tax reporting and financial analysis.

​
Splitting Complex Transactions

Some transactions need to be split across multiple categories or business segments:
  1. Get the full transaction details to understand the total amount
  2. Update with split information, specifying amounts and categories for each split
  3. Add notes documenting the reason for the split
Splits are common for transactions like business trips (split between travel, meals, and lodging) or bulk purchases from retailers like Amazon (split by category: office supplies, equipment, etc.).

​
Examples

​
”What transactions need my review?”

AI uses: query_bank_transactions with:
  • reviewStatus: “needs review”
AI responds: “You have 8 transactions that need your review. Here are the highlights:
  • $1,245.00 from ‘TECH SOLUTIONS INC’ on January 10th
  • $687.50 from ‘AMAZON.COM’ on January 9th
  • $450.00 from ‘OFFICE DEPOT’ on January 8th
Would you like me to walk through these one by one?"

​
"Can you recategorize this transaction as Business Meals?”

AI uses: update_bank_transaction with:
  • category: “Business Meals”
AI responds: “I’ve recategorized this transaction as Business Meals. The $67.43 expense from ‘OLIVE GARDEN’ on January 11th is now properly categorized in your books.”
The AI should identify which transaction the user is referring to from context in the conversation. If ambiguous, the AI should use query_bank_transactions to find the most recent transaction or ask for clarification.

​
”What was my most recent transaction from Ace Hardware?”

AI uses: query_bank_transactions with:
  • vendor: “Ace Hardware” or description: “Ace Hardware”
  • Sort by date (most recent first)
  • Limit to 1 result
AI responds: “Your most recent transaction from Ace Hardware was $124.87 on January 9th. It’s currently categorized as ‘Repairs & Maintenance’. The bank description shows ‘ACE HARDWARE #2847’. Would you like to see more details or update the categorization?"

​
"Can you find all transactions similar to the one I just edited?”

AI uses: Multiple query_bank_transactions calls based on the recently edited transaction’s attributes:
  • Same vendor
  • Similar amount range (±20%)
  • Same or similar description pattern
  • Within a relevant time window
AI responds: “I found 5 transactions similar to the one you just edited (the $89.00 ‘SHELL GAS STATION’ transaction you categorized as ‘Vehicle Expenses’):
  1. $92.50 from ‘SHELL GAS STATION’ on January 8th - currently uncategorized
  2. $85.00 from ‘SHELL GAS STATION’ on January 5th - currently uncategorized
  3. $78.45 from ‘SHELL GAS STATION’ on December 29th - categorized as ‘Auto & Travel’
  4. $95.20 from ‘CHEVRON’ on January 7th - currently uncategorized
  5. $88.00 from ‘BP GAS’ on January 4th - currently uncategorized
Would you like me to categorize all the uncategorized gas station purchases as ‘Vehicle Expenses’?”