Skip to content

SciAgentsDiscovery

Automate research hypotheses with multi-agent graph reasoning

Visit official site

What SciAgentsDiscovery is

SciAgentsDiscovery is a system that automates scientific research with a multi-agent system and a large-scale ontology knowledge graph. It integrates large language models, data retrieval tools and multi-agent learning to generate and refine research hypotheses.

What you can do with it

  • Organise and connect scientific concepts with an ontology knowledge graph
  • Combine large language models with data retrieval tools
  • Generate and refine research hypotheses automatically
  • Reveal latent mechanisms and material design principles

Application scenarios

  • Link the keywords silk and energy intensive to propose a new hypothesis
  • Let generated hypotheses guide further scientific exploration
  • Use detailed documents for insight into material design and properties

Getting started

  • Install the required GraphReasoning package and API
  • Clone the SciAgentsDiscovery repository from GitHub
  • Run the Jupyter notebooks under the Notebooks directory
  • Choose a non-automated or automated multi-agent framework

Pros & cons

✓ What we like

  • Concepts are organised through a knowledge graph
  • Detailed documents are produced alongside hypotheses

! What to watch out for

  • The workflow needs the GraphReasoning package and an API
  • The framework is chosen between non-automated and automated modes

FAQ

What does the system produce?

The material lists research hypotheses, revealed mechanisms and design principles, and detailed documents.

How do I run it?

The steps name the GraphReasoning package and API, cloning the GitHub repository and running the Jupyter notebooks under Notebooks.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review