Skip to main content

​
MCP Tools

Haptic provides three MCP tools for accessing your financial data. Each tool is designed for specific use cases and returns structured data your AI can interpret.

​
Tool Overview

get_account_summary

Use for: Net worth, balances, financial snapshotReturns comprehensive financial overview with categorized assets and liabilities.

get_transactions

Use for: Historical spending, custom date rangesQuery transactions with flexible date filtering.

get_recent_transactions

Use for: Latest activity, quick checksGet last 24 hours of transactions across all accounts.

​
get_account_summary

Get a comprehensive financial summary including account balances, net worth, assets, and liabilities breakdown.

​
Parameters

None - automatically uses the authenticated user’s accounts.

​
Use Cases

“What’s my current net worth?”Returns total assets minus total liabilities with full breakdown.
“Show me all my account balances”Lists all connected accounts with current balances categorized by type.
“How much do I have in investments?”Breaks down assets into Cash, Investments (retirement & brokerage), and other categories.
“What’s my total credit card debt?”Summarizes all liabilities including credit cards and loans.

​
Response Format

The tool returns a formatted text summary optimized for different account compositions: Example 1: Mixed Accounts 
Financial Summary:

NET WORTH: $45,230.15

ASSETS ($52,000.00):
Cash:
  - Chase Checking: $8,500.00
  - Ally Savings: $15,000.00

Investments:
  - Vanguard 401(k): $28,500.00

LIABILITIES ($6,769.85):
Credit:
  - Chase Sapphire: $4,200.50
  - Amex Blue Cash: $2,569.35
Example 2: Credit Cards Only 
Credit Card Debt Summary:

Total Debt: $6,769.85

Accounts:
  - Chase Sapphire: $4,200.50
  - Amex Blue Cash: $2,569.35
The tool intelligently formats output based on your account types - no negative net worth shown if you only have credit cards.

​
Data Updated

This tool triggers a Plaid API call, so data is real-time (within seconds of your bank’s last update).

​
get_transactions

Get transactions for a custom date range across all connected accounts.

​
Parameters

​
start_date
string
Start date in YYYY-MM-DD format (e.g., “2024-01-15”)Default: 24 hours ago if omitted
​
end_date
string
End date in YYYY-MM-DD format (e.g., “2024-01-31”)Default: Today if omitted

​
Natural Language Support

Your AI will automatically convert natural language to proper date formats:
You sayAI converts to
”last week”start_date: 7 days ago, end_date: today
”this month”start_date: 1st of current month, end_date: today
”last 30 days”start_date: 30 days ago, end_date: today
”January 2024”start_date: “2024-01-01”, end_date: “2024-01-31”

​
Use Cases

“What did I spend in January?”AI calls: get_transactions({ start_date: "2024-01-01", end_date: "2024-01-31" })
“Show me all my restaurant transactions from last week”AI calls get_transactions with last week’s dates, then filters by category.

​
Response Format

Transactions (2024-01-01 to 2024-01-31):

2024-01-15 | Chase Checking | Whole Foods | $125.43 (debit) | Groceries
2024-01-14 | Chase Sapphire | Netflix | $15.99 (debit) | Entertainment
2024-01-12 | Chase Checking | SALARY DEPOSIT | $5,000.00 (credit) | Income

Total: 87 transactions

​
Limits

  • Maximum range: 2 years
  • Format: YYYY-MM-DD required
  • Validation: Start date must be before end date
Date format must be exactly YYYY-MM-DD. Invalid formats will return an error with clear instructions.

​
get_recent_transactions

Get transactions from the last 24 hours across all connected accounts.

​
Parameters

None - automatically fetches last 24 hours.

​
Use Cases

“What did I spend yesterday?”Quick overview of recent activity without specifying dates.
“Did my Netflix payment go through?”Check for recent transactions from specific merchants.
“Any unusual transactions today?”AI can quickly scan for suspicious or unexpected charges.

​
Response Format

Recent Transactions (Last 24 Hours):

2024-12-15 | Chase Checking | Starbucks | $6.75 (debit) | Restaurants
2024-12-15 | Amex Blue Cash | Amazon | $42.99 (debit) | Shopping

Total: 2 transactions
This tool is optimized for speed - use it for quick checks instead of the full get_transactions tool when you only need recent activity.

​
Common Patterns

​
Categorized Spending

Your AI can call get_transactions and then group by category: 
You: "How much did I spend on food last month?"

AI:
1. Calls get_transactions for last month
2. Filters for Food & Drink categories
3. Sums the amounts
4. Responds: "You spent $450 on food last month"

​
Balance Changes

Track how balances change over time: 
You: "Did my net worth increase this month?"

AI:
1. Calls get_account_summary (current)
2. Calls get_transactions for the month
3. Calculates starting balance from transactions
4. Compares and reports the change

​
Spending Alerts

Set up conversational alerts: 
You: "Alert me if I spend more than $100 at Target"

AI:
1. Periodically calls get_recent_transactions
2. Filters for Target transactions
3. Checks amounts
4. Alerts if threshold exceeded

​
Tool Performance

All tools are real-time - data is fetched directly from Plaid when requested:
  • Latency: Typically 1-3 seconds per call
  • Accuracy: Matches your bank’s latest data
  • Caching: Recommended for repeated queries within short timeframes
The last_updated timestamp for your accounts is updated each time these tools are called successfully.

​
Error Handling

Tools return clear error messages for common issues:
ErrorMeaningSolution
”No connected accounts found”No Plaid accounts linkedConnect a bank account in your dashboard
”Invalid date format”Wrong date format usedUse YYYY-MM-DD format
”No subscription”Subscription requiredSubscribe at haptic.sh/dashboard
”No transactions found”No activity in date rangeTry a different date range

​
Next Steps

API Reference

Technical tool specifications

Connection Guide

Set up MCP in your AI

Security

How your data is protected

Troubleshooting

Common issues and fixes