liteparse

LiteParse is a local document parsing skill designed to handle a wide variety of unstructured file formats, including PDFs, DOCX, PPTX, XLSX, and images. It allows users to extract text, perform multi-format conversions, and capture spatial layout information without relying on cloud services or large language models. By providing a local solution, LiteParse addresses common challenges such as privacy concerns, slow upload/download times, and dependency on internet connectivity, making it particularly useful for sensitive or large-scale document processing tasks.

The skill offers robust features such as single-file parsing, batch directory parsing, and page-specific screenshot generation. Users can specify output formats (JSON or plain text), apply OCR for image-based content, control page ranges, and adjust rendering quality through DPI settings. It supports external OCR servers for enhanced accuracy, and provides options to optimize performance by disabling precise bounding boxes or skipping diagonal text. Additionally, LiteParse integrates seamlessly with command-line workflows using the `lit` CLI, allowing automated scripts and customizable TypeScript scripts for more complex processing tasks.

LiteParse is ideal for developers, data analysts, and researchers who need to extract structured data from documents for analysis, machine learning, or automated reporting. It is suitable for environments where cloud services are unavailable or undesired, and where speed, flexibility, and local processing are essential. Typical use cases include extracting tables from PDFs, converting office documents to structured JSON, generating page screenshots for AI agents, and batch-processing large directories of documents while maintaining control over OCR and layout fidelity.