v0.22.44 Technology Demonstrator Python 3.11+ Win / Mac / Linux

Mission & Concept

OpenData Tool is a technology demonstrator designed to evaluate if AI can reliably interpret research directories and draft high-quality metadata records with minimal user intervention.

Current State: This is a working prototype and proof-of-concept for automated RODBUK packaging, not a finished production product.

Read-Only Safety

Analysis is strictly non-destructive. The tool never modifies your original research files.

Expert Extraction

Specialized logic for LaTeX, DocX, PDF, HDF5, and DICOM (more formats in development).

Interactive Chat

Refine suggestions through simple dialogue. The AI learns from your clarifications in real-time.

Project Workflow

1

Guided Setup

Initial language selection and AI capability configuration (Google OAuth2). Settings are persisted globally.

2

Analysis Loop

Point to a directory. The agent injects field protocols and enters an iterative chat to propose metadata.

3

Refinement

Knowledge capture of lab-specific rules and final generation of the RODBUK-ready metadata package.

Documentation

Welcome to the Beta Portal

The OpenData Tool is currently in an active testing phase. This portal serves as your central resource for understanding, installing, and contributing to the project.

Quick Start

If you're ready to dive in, the Installation Guide will help you get the tool running on Windows, Linux, or macOS in minutes.

Learn the Basics

New to metadata preparation? Read our Beta Tester Manual to understand how to collaborate with the AI assistant.

For Developers

Interested in how the tool works under the hood? Our Developer Manual covers the architecture, build system, and how to write custom extractors.

Installation

Ready for deployment on all major platforms.

2. PyPI Package

uvx opendata-tool fastest
pipx install opendata-tool isolated

3. Development

$git clone https://github.com/jochym/opendata.git

$cd opendata

$pip install -e ".[dev]"

$python src/opendata/main.py

Recommended for contributors and researchers with Python experience.