GenAI Expense Tracker

⚠️ EXPLORATION NOTICE: This project exists solely for prompt engineering exploration and educational purposes. While functional, this approach to expense tracking is not necessarily optimal or production-ready. There are numerous dedicated financial applications, spreadsheet solutions, and purpose-built tools that provide more robust, secure, and feature-rich expense tracking capabilities. This repository demonstrates what's possible with LLM prompting rather than what's practical for daily expense tracking.

Overview

This repository documents the iterative development and refinement of an Expense Tracking Assistant prompt for Large Language Models (LLMs). The primary focus is tracking the evolution of the Prompt.md file as it's optimized for better performance, clarity, and functionality.

Purpose

The main purposes of this repository are:

2. LLM Behavior Analysis: Observe how different prompt formulations affect the assistant's responses, capabilities and limits
3. Educational Resource: Serve as a reference for initial prompt engineering
4. Performance Optimization: Document the process of refining prompts to achieve more accurate budget tracking and expense management

Core Components

• Prompt.md: The central file containing the system prompt for the Expense Tracking Assistant
• Change History: Git commit history showing the evolution of prompt design decisions
• Testing Examples: Sample interactions demonstrating the assistant's capabilities at different stages (TODO)

The Expense Tracking Assistant

The assistant defined in Prompt.md is designed to:

• Manage a personal expense ledger in Markdown table format
• Process expense transactions based on natural language commands
• Track spending against predefined budget categories
• Provide alerts when spending exceeds budget limits
• Maintain accurate expense records with clear formatting

Getting Started

To experiment with the prompt:

1. Review the current version of Prompt.md
2. Use the prompt with an LLM that supports system prompts (e.g., ChatGPT, Claude, etc.)
3. Test the assistant with sample expense transactions
4. Observe the assistant's responses, particularly its ledger management and budget tracking

Contributing

Contributions to improve the prompt are welcome. When suggesting changes, please:

1. Explain the reasoning behind your proposed modifications
2. Provide examples of how the changes improve the assistant's performance
3. Consider edge cases and potential limitations

License

This project is open source and available under the MIT License.