Liteparse
liteparse is an open-source document parser from the LlamaIndex (run-llama) team, designed for fast, structure-preserving conversion of documents into formats suitable for LLM ingestion pipelines. It handles PDFs, Word documents, HTML, and other common formats while retaining structural elements — tables, headings, lists — that simpler text extractors discard.
The tool is built as a direct component for RAG and agent workflows, with the design goal of being faster and more reliable on difficult formats (scanned PDFs, multi-column layouts) than generic extraction libraries.
Placed in Tools > Assess at inner position because document parsing is a foundational step in any LLM pipeline that works with real-world documents, and the quality of parsing directly affects downstream retrieval and generation quality. liteparse's provenance from the LlamaIndex team gives it credible integration with the broader LLM tooling ecosystem. Worth assessing for any pipeline where document ingestion quality is a bottleneck.
