invoice-processor

Invoice extraction automates the process of pulling structured data like invoice number, date, buyer/seller names, and amounts from PDF, JPG, JPEG, and PNG files using AI vision recognition. This Skill processes invoices through OCR, converts PDFs to images at 200 DPI, encodes them in base64, and sends them to a cloud vision model for field identification.

Yes, batch processing supports up to five invoices concurrently. The Skill processes multiple files in parallel, tracks status for each, and consolidates results into a single JSON output before converting to Excel, reducing processing time for large invoice volumes.

The Skill extracts 12 standardized fields: invoice type, number, date, buyer name, seller name, total excluding tax, tax amount, total including tax, and item names. Output is generated as formatted JSON for downstream processing or auditing before Excel export.

Run the environment check script to verify dependencies (aiohttp, PyMuPDF, openpyxl), then execute the OCR processing script pointing to your invoice folder, and convert the resulting JSON to Excel. Quick Start commands are provided for each step with configurable input/output paths.

The Skill requires API key configuration for the cloud vision model and performs local environment checks before processing. Images are base64 encoded and sent to the vision API; guidance on privacy settings and local configuration is included to ensure secure handling of invoice data.

The Skill processes JPG, JPEG, PNG, and PDF formats. Language support depends on the underlying cloud vision model's capabilities; the Skill handles format conversion and standardization but does not apply language-specific filtering or translation.