YNAB Statement Formatter

Converts credit card statement images to YNAB-compatible CSV format using AI-powered extraction.

Features

• AI-powered transaction extraction from statement images
• Multi-provider support: Google AI Studio, OpenRouter, z.ai, LM Studio
• Model Testing Lab to compare AI performance
• Smart date parsing for various formats
• Batch processing of multiple statements
• Custom instructions for bank-specific formats
• Editable results before export

Quick Start

1. Clone and install:

git clone https://github.com/yjsoon/ynab-llm-formatter.git
cd ynab-formatter
npm install

2. Configure API key (Google AI Studio recommended):

cp .env.example .env.local
# Edit .env.local with your API key

3. Start the server:

npm run dev

4. Open http://localhost:3000

Configuration

Option 1: Google AI Studio (Default)

LLM_PROVIDER=googleaistudio
GOOGLEAISTUDIO_API_KEY=your_api_key_here
GOOGLEAISTUDIO_MODEL=gemini-2.5-flash-lite

Get API key from Google AI Studio.

Option 2: OpenRouter

LLM_PROVIDER=openrouter
OPENROUTER_API_KEY=your_api_key_here
OPENROUTER_MODEL=google/gemini-2.5-flash-lite

Sign up at openrouter.ai for API access.

Option 3: z.ai

LLM_PROVIDER=z.ai
Z_AI_API_KEY=your_api_key_here

Option 4: LM Studio (Local)

LLM_PROVIDER=lm-studio
LM_STUDIO_URL=http://localhost:1234/v1
LM_STUDIO_MODEL=qwen2.5-vl-7b-instruct

Download LM Studio and load a vision-capable model.

Usage

1. Select AI model from dropdown
2. Upload statement images (PNG, JPG, WebP)
3. Add custom instructions if needed
4. Process and review extracted transactions
5. Download CSV for YNAB import

Custom Instructions

• "Dates are in DD/MM/YYYY format"
• "Ignore transactions marked as REVERSAL"
• "Foreign currency amounts shown as (USD 50.00)"
• "Statement is from DBS Bank Singapore"

Model Testing Lab

Visit /test to compare AI models:

• Test extraction accuracy across providers
• Compare processing speeds and costs
• Find the best model for your statement format

Development

npm run dev    # Development server
npm run build  # Production build
npm run start  # Production server
npm run lint   # ESLint

License

MIT