Verified project record

xberg-io/xberg

xberg-io/xberg is a polyglot document extraction engine with 15 language bindings and a unified Rust core that extracts text, tables, metadata, and structured information from over 97 formats. It provides multiple interfaces including a CLI, library, REST API server, and an MCP server, with optional enrichment services that can rely on local or remote models.

Project overview

The project consolidates fragmented document processing libraries into a single engine, supporting formats like PDFs, Office documents, and images, while eliminating the need for manual OCR setup or format detection.

Project type: MCP · RAG
Use cases: Knowledge Q&A
Deployment: Refer to project documentation
License: MIT

Best for

Developers, AI engineers, and data teams who need to extract text, tables, and metadata from diverse document formats into structured representations.

Key capabilities

Extracts clean text, tables, and metadata from 97+ formats including PDFs, Office documents, images, HTML, email, and archives.
Runs as an MCP server exposing 9 extraction tools, 3 prompts, and 4 resources for IDE agents like Claude or Cursor.
Runs as a REST API server handling all formats via a single POST endpoint and streaming large files.
Performs NER, redaction, summarization, translation, QR code detection, page classification, and keyword extraction.

Limitations and risks

Functions like provider-hosted embeddings, structured extraction, and VLM OCR rely on optional external services including OpenAI, Anthropic, or Google.
Telemetry configuration is not documented.

Getting started

Installation is rated as easy, offering one-line commands like 'pip install xberg' or 'brew install xberg-io/tap/xberg'. Users can extract text via CLI using 'xberg extract document.pdf'.

Alternatives and comparisons

Provides advanced PDF understanding, unified document representation, and integrations with the generative AI ecosystem.
Turns unstructured documents into structured JSON using LLM-driven extraction defined by natural language prompts.
An open-source document intelligence platform that builds a programmable citation graph using structured extraction and AI agents.

Project comparisons

Evidence and sources

Open original GitHub project