cli-pdf-extract

A fast Rust CLI wrapper around pdf_oxide that lets LLMs peek into PDFs without paying the cost of loading and synthesizing whole documents. It extracts page text as Markdown for quick ingestion and can also pull only highlights (plus notes), which is ideal when you want the essence without the full context and need higher LLM throughput.

Features

• Extract pages as Markdown (single page, range, or all pages by default)
• Extract only highlight annotations and their notes (--highlight)
• Write to stdout (for piping) or a file (--output)
• Designed for low-latency LLM workflows (fast “peek” into large PDFs)

Prerequisites

• Rust and Cargo installed (rustup recommended)

Installation

Option 1: Build and run locally

cargo run -- <PDF_PATH> --page 0

Option 2: Install as a local CLI binary

From the repository root:

cargo install --path .

Then run:

cli-pdf-extract <PDF_PATH> --page 0

Usage

Show help

cli-pdf-extract --help

Single page extraction

cli-pdf-extract examples/main.pdf --page 0 --output extract.md

Page range extraction (inclusive)

cli-pdf-extract examples/main.pdf --start-page 0 --end-page 5 --output extract.md

Extract highlights only (fastest LLM pass)

cli-pdf-extract examples/main.pdf --highlight

Pipe directly to another command / LLM tool

cli-pdf-extract examples/main.pdf --start-page 0 --end-page 5 | cat

Notes

• Pages are zero-indexed.
• --start-page and --end-page must be provided together.
• --page cannot be combined with range flags.
• Pro-tip: add standardized tags to annotation notes (e.g., <problem-simulations>, <paper-idea>) to enable downstream clustering, trend discovery, and routing.

License

MIT. See LICENSE.

Author

Edgar Torres (edgar.torres@ki.uni-stuttgart.de)