The open-source alternative to Claude Science,for research automation.
Runcell Science is an open-source science agent for reading papers, writing code, planning experiments, and running domain skills across BioTech, chemistry, and data-heavy research. It runs locally on Claude Code and Codex, so your research workflow stays inspectable and under your control.
$ git clone https://github.com/runcell-ai/runcell-science.gitAn agent for science work
Scientists do not need another general chat window. They need an agent that can move through the daily research loop: understand the literature, operate on data and code, use domain tools, and leave behind artifacts that can be inspected and reused.
Skills for BioTech and chemistry
Runcell Science is built around domain skills: chemistry structure work, bioinformatics routines, literature workflows, experimental notes, and code-backed analysis. The point is to give scientists an agent that understands the shape of scientific work.
Research automation, end to end
Use the agent to collect context, inspect files, run analyses, update notebooks, draft summaries, and save the resulting artifacts back into the session. Repeated scientific chores become reproducible agent runs instead of one-off chat transcripts.
Papers, experiments, and code in one loop
A science agent should read papers, extract methods, write draft sections, prepare experiment plans, generate code, run checks, and keep the evidence attached. Runcell Science keeps those steps in one local workspace.
“The product is the science workflow: skills, sessions, artifacts, and repeatable research automation.”
What the agent helps you do
Research moves across papers, protocols, data, notebooks, molecules, and experiments. Runcell Science keeps that loop in one workspace, so the agent can do useful scientific work rather than only answer isolated questions.
- 2.1
Read papers and extract methods
Ask the agent to digest scientific papers, pull out protocols, summarize assumptions, compare methods, and turn findings into notes you can reuse in the next step of the project.
- 2.2
BioTech and chemistry skills
Skills package repeatable domain workflows for the agent, from molecule and assay work to bioinformatics and data analysis. The library is early, but the extension point is open and central to the product.
- 2.3
Write and run research code
Let the agent inspect data, write scripts, run checks, update notebooks, generate figures, and explain the result. Code execution and generated files stay attached to the research session.
- 2.4
Draft papers and experiment notes
Move from raw outputs to structured writing: method summaries, result narratives, experiment plans, caveats, and follow-up questions. The agent keeps the chain of work visible.
- 2.5
Artifacts stay with the work
Figures, notebooks, molecule edits, and files the agent produces are captured as artifacts on the session and rendered in place, instead of disappearing into terminal scrollback.
- 2.6
Local-first and open source
Runs on your machine, in your browser, against your project files. Apache-2.0 licensed and self-hostable, so research teams can inspect and adapt the system before trusting it.
An open alternative to Claude Science
Claude Science points at the right destination: an agent that can help with scientific work. Runcell Science takes that idea into an open, local-first project that teams can inspect, extend, and run with the tools they already use.
Table 1. Architectural comparison across the three approaches.
| Dimension | Runcell Science | Self-built harness agents | Claude Science (official) |
|---|---|---|---|
| Agent harness | Claude Code & Codex | Custom-built | Anthropic internal |
| Reuses your existing Claude Code / Codex subscription | Yes | No — separate API keys | No |
| Open source / self-hostable | Yes — Apache-2.0 | Yes | No |
| Cross-vendor (Claude + Codex) | Yes | Varies | No |
| Skills library size | Growing | Often larger | Large |
Note. Yes — others currently ship more skills. Skills are the easiest layer to grow in the open; the harness is the hardest to rebuild. That trade is the point.
First session in five minutes
$ git clone https://github.com/runcell-ai/runcell-science.git
$ cd runcell-science
$ ./scripts/dev.shhttp://127.0.0.1:27183$ yarn install
$ yarn devFrequently asked 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?
Yes. Runcell Science drives the Claude Code CLI you already have installed and signed in. If your Claude plan covers Claude Code, it covers Runcell Science sessions too. The same applies to Codex and your existing OpenAI plan.
Do I need my own API keys?
No. Runcell Science does not proxy your requests through a hosted service or bill you per token. It runs on top of the local claude or codex CLI, authenticated with the account you are already signed in to, so there is no separate API-key billing.
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?
A native Windows desktop build is on the roadmap — it is a gap in this category we intend to fill. Today Runcell Science runs as a local web app; on Windows, running it under WSL is the practical path. Track progress in the GitHub repository.
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.
Read the code. Run the science.
© 2026 runcell.dev (Kanaries Data Inc.) · Apache-2.0