unstract

unstract

LLM-based tool for extracting structured data from unstructured documents, supporting API deployment and ETL workflows.

MCPDataOpen source
Type
MCP
Transport
http
Open source
Yes
GitHub Stars
★ 6.7k
Source
mcp-github

Overview

Unstract uses large language models to extract structured JSON data from documents (such as PDFs, images, etc.). Users can define what to extract using natural language prompts and deploy the extraction pipeline as an API or ETL process. Suitable for finance, insurance, healthcare, KYC/compliance, and other domains. Define extraction patterns via Prompt Studio and process new document types in minutes. Supports multiple LLM providers (e.g., OpenAI, Anthropic) and can connect to AI agents via the MCP protocol.

Capabilities

  • Define extraction patterns using natural language
  • Rapid processing of new document types
  • Support for multiple LLM providers
  • API deployment and ETL pipelines
  • Connect to AI agents via MCP protocol

Use cases

Automated financial document processingInsurance claim document parsingStructured medical record extractionCompliance checks

Setup

Requires: Docker & Docker Compose8 GB RAMLinux或macOS
git clone https://github.com/Zipstack/unstract.git; cd unstract; ./run-platform.sh

This information was compiled by AI from public sources and may contain inaccuracies — please refer to the source.

FAQ

How do I get started with Unstract?

After cloning the repository, run ./run-platform.sh and access http://frontend.unstract.localhost to log in.

What operating systems are supported?

Supports Linux and macOS (Intel or M-series).

Related skills