How does the skill handle small price differences between receipts and bank records?

The skill applies a ±1% tolerance rule on the total amount to account for common discrepancies like currency conversion micro-adjustments, rounding, or tip additions.

Can I link multiple receipts to a single bank transaction?

Yes. While the skill warns you if a transaction is already linked, it allows for multiple links to accommodate split payments or multiple purchases that sum to a single bank entry.

What happens if a transaction posts several days after the purchase?

The logic includes a ±1 day window for standard receipts and a broader 5-day window for payslips to account for weekend delays and bank processing times.

Does this skill support foreign currency matching?

Yes, it compares the receipt data against the converted local currency amount (e.g., DKK) found in the bank transaction, though it flags these with lower confidence scores for user review.

Financial Transaction & Receipt Matching

Name: Financial Transaction & Receipt Matching
Author: peerjakobsen

bypeerjakobsen

•

Productivity & Workflow

Automates the reconciliation of uploaded receipts and payslips with bank transactions using multi-signal fuzzy matching and confidence scoring.

The Transaction Matching skill provides a robust framework for linking financial documents to bank records within the SmartSpender repository. It employs a sophisticated logic engine that evaluates amount tolerances, date windows, and merchant name normalization to ensure accurate reconciliation. By calculating confidence scores for every potential match, it enables seamless automated linking for clear matches while providing a guided confirmation workflow for ambiguous entries, significantly reducing the manual effort required for personal or business expense tracking.

Key Features

01Dynamic date windowing (±1 day for receipts, 5 days for payslips) to handle bank posting delays.

02Merchant name normalization and pattern matching against transaction descriptions.

03Weighted confidence scoring for automated 'Auto-link' or 'User-confirmation' decisions.

04Detection of existing matches to prevent duplicate document linking.

052 GitHub stars

06Fuzzy amount matching with ±1% tolerance to account for tips and rounding.

Use Cases

01Verifying monthly salary deposits against digital payslips using income patterns.

02Cleaning and categorizing financial data by linking merchant metadata to generic bank descriptions.

03Automating the reconciliation of retail receipts with bank statement CSV exports.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add peerjakobsen/smartspender transaction-matching

For use in Claude.ai and ChatGPT

Download Skill