An open-source project

Apache-2.0

The open-source alternative to Claude Science,for research automation.

Read papers, run analysis code, and inspect scientific artifacts in one local workspace. Built on Claude Code and Codex, with source you can explore and skills you can extend.

Open source · Local workspace · Your choice of agent

01 / A research workflow

From a question to work you can inspect.

Ask → Run → Review
Notebook workflow / clustering analysis
Workflow preview recorded in the Runcell JupyterLab extension. The Science workspace is shown below.
  1. 01

    Start with a question

    Bring your dataset, notebook, or research context into the conversation.

  2. 02

    Work through the analysis

    Follow the code and tool activity as the agent works through each step.

  3. 03

    Inspect. Revise. Continue.

    Review the outputs, question the method, and carry the files into your next experiment.

§ 2 / Runcell Science

Keep the research together.

Papers, experiments, code, and the files they produce belong in the same working context.

Explore the project
Runcell Science: a medicinal chemistry conversation beside a Ketcher molecule editor artifact, with Codex selected as the agent.
The Science workspace: conversation, tool activity, and a chemistry artifact.Click image to enlarge ↗

Read papers. Keep the methods.

Compare approaches, extract protocols, and turn literature into working notes for the next step of your research.

Run code. Inspect the evidence.

Work with scripts and notebooks, review generated figures, and keep the resulting files beside the session.

Bring your domain tools.

Use and extend skills for BioTech, chemistry, and data analysis. The library is growing, and the workflow is yours to adapt.

Two ways to work with Runcell

Choose the workspace that fits.

Open-source project

Runcell Science

For researchers and developers who want to inspect the source, adapt scientific skills, and shape their own local research workspace.

Apache-2.0 · Local setup · No Runcell account required

Set up Runcell Science

From the Runcell team

Runcell Desktop

Prefer a downloadable app for your notebook workflow? Explore Runcell Desktop, our separate product for working with AI in Jupyter notebooks.

Desktop beta · See available builds on the download page

Explore Runcell Desktop

§ 3 / Runcell Science

An open alternative to Claude Science.

Choose based on how you want to work: your agent runtime, your tools, and how much of the workspace you want to adapt.

Runcell Science and Claude Science: runtime, customization, access, and workspace.
What mattersRuncell ScienceClaude Science
Agent runtimeClaude Code or CodexClaude
Source & customizationApache-2.0 source. Modify the workspace and add your own skills.Supports custom skills and connectors.
Getting accessRun the source locally with a configured agent CLI. Provider access and usage limits apply.Available to Claude Pro, Max, Team, and Enterprise users. Admin enablement may be required.
Where you workA local workspace for project files, sessions, and scientific artifacts.macOS, Linux, or remote compute through SSH and HPC workflows.

Swipe the table to compare both workspaces.

An independent project by the Runcell team, not affiliated with Anthropic. Claude Science details checked September 29, 2026 against Anthropic’s product announcement. Features and access may change.

§ 4 / Runcell Science

Start your first local session.

Clone the project, start the workspace, and bring a research question of your own.

1. Clone and start
git clone https://github.com/runcell-ai/runcell-science.git
cd runcell-science
./scripts/dev.sh
2. Open in your browser
http://127.0.0.1:27183
Prefer to start services manually?
From the cloned repository
yarn install
yarn dev
Read the setup guide on GitHub

§ A / Runcell Science

A few practical questions.

Is there an open-source alternative to Claude Science?

Yes. Runcell Science is an open-source alternative to Claude Science for research automation. It helps with papers, experiments, code, and science workflows while running locally on your machine under the Apache-2.0 license.

Does Runcell Science work with my existing Claude Code subscription?

Runcell Science uses the Claude Code or Codex CLI installed on your machine. Sign in through the CLI with an account that has access to it. Your provider’s plan, usage limits, and billing settings still apply.

Do I need my own API keys?

For the Claude Code or Codex setup shown here, use your signed-in CLI; there is no additional Runcell API key to configure. Authentication and any model charges depend on how you have configured that CLI.

How is this different from building an agent with its own harness?

Runcell Science focuses on the science workflow layer: research sessions, artifacts, and domain skills for papers, experiments, code, BioTech, and chemistry. It uses Claude Code and Codex for the underlying agent runtime so more effort can go into scientific workflows.

Is Windows supported?

For the local web workflow on Windows, use WSL and set up the project and your chosen agent CLI there. Desktop packaging is evolving; check the GitHub repository for current build availability and platform requirements.

How does Runcell Science relate to Runcell?

Runcell Science is an open-source project by the Runcell team. Runcell itself is a separate commercial product — an AI agent for Jupyter notebooks. Runcell Science is independent and Apache-2.0 licensed; you do not need a Runcell account to use it.

Your next research session

Read the code. Run the science.

Explore on GitHub