transaction-history

name	transaction-history
description	Retrieve recent transaction history for a customer account with filtering, search, and export — uses Faker for simulated data
enabled	true

Instructions

When the user asks to see recent transactions, account activity, statement, or spending history, use this skill.

1. Generate Transaction Data

This skill uses the Faker MCP server (configured in .mcp.json) to generate realistic transaction history. Call Faker MCP tools to produce individual data points, then assemble them into transactions.

Faker MCP tools to use:

• faker_date_between — generate transaction dates within the last 90 days
• faker_random_element — pick merchants, categories, and transaction statuses
• faker_pyfloat — generate transaction amounts
• faker_bothify — generate transaction reference numbers (e.g., "TXN-########")

Transaction categories and merchants:

Category	Example Merchants
Groceries	Whole Foods, Trader Joe's, Kroger, Safeway, Costco
Dining	Starbucks, Chipotle, McDonald's, Olive Garden, DoorDash
Transportation	Uber, Lyft, Shell Gas, BP, EZ-Pass
Shopping	Amazon, Target, Walmart, Best Buy, Nordstrom
Utilities	ConEdison, AT&T, Comcast, Water Authority, National Grid
Healthcare	CVS Pharmacy, Walgreens, Dr. Smith Office, LabCorp
Entertainment	Netflix, Spotify, AMC Theatres, Apple iTunes
Transfer	Zelle Transfer, Wire Transfer, ACH Transfer, Venmo
Income	Direct Deposit - Employer, ACH Credit, Interest Payment
Fees	Monthly Service Fee, ATM Fee, Overdraft Fee

Example flow:

1. Use Faker MCP to generate dates, merchant names, and amounts
2. Use code_interpreter to assemble into a sorted transaction list and apply filters
3. Present as a formatted table

Expected output per transaction:

{
  "date": "2026-03-18",
  "description": "Whole Foods",
  "category": "Groceries",
  "amount": -87.43,
  "type": "Debit",
  "status": "Posted",
  "reference": "TXN-48219753"
}

2. Response Format

Present transactions in a clean table:

Date	Description	Category	Amount	Status
2026-03-18	Whole Foods	Groceries	-$87.43	Posted
2026-03-18	Direct Deposit	Income	+$3,250.00	Posted
2026-03-17	Uber	Transportation	-$24.50	Posted
2026-03-16	Netflix	Entertainment	-$15.99	Pending

3. Filtering & Search

Support user requests to filter transactions by:

• Date range: "Show me transactions from last week"
• Category: "Show me all dining expenses"
• Amount range: "Transactions over $100"
• Merchant: "Search for Amazon"
• Type: "Show only credits/deposits"

4. Spending Summary

If the user asks for a spending overview, aggregate by category:

import pandas as pd

df = pd.DataFrame(transactions)
summary = df[df["amount"] < 0].groupby("category")["amount"].agg(["sum", "count"])
summary.columns = ["Total Spent", "# Transactions"]
summary = summary.sort_values("Total Spent")
print(summary)

5. Export

If the user wants a statement or export:

• Generate a CSV file to /tmp/transactions_YYYYMMDD.csv
• Reference the file path for download via file-sharing

Chaining

• Faker MCP — generates realistic transaction data points
• code_interpreter — assembles data, performs aggregations, filtering
• data-analysis — deeper spending analytics and charts
• file-sharing — export transaction data as CSV
• account-lookup — user often checks balance before reviewing transactions

Constraints

• Maximum 90 days of transaction history available. If the user requests transactions older than 90 days, inform them that only the last 90 days are available online and suggest they visit a branch or contact customer service at 1-800-OLYMPUS to request a paper statement.
• Always show most recent transactions first
• Mask account numbers in exports (show only last 4 digits)

Constraints

• All transactions are simulated via Faker
• Always show most recent transactions first
• Maximum 90 days of history per request
• Mask account numbers in any exported files