Jupyter Express

AI-powered Python coding agent for Jupyter Notebook/Lab.

Jupyter Express is an AI-powered and context-aware Python coding agent that lives 𝘪𝘯𝘴𝘪𝘥𝘦 your Jupyter Notebook/Lab. Type a prompt, get ready-to-run Python code inserted into a new cell. No tab switching, no copy-pasting code from generic AI

🎉 New in 1.1.0: Jupyter Express now works in JupyterLab and Jupyter Notebook 7, in addition to Jupyter Notebook 6.

---

Installation

Open a terminal and run the following command:

pip install jupyter-express

Important: Because Jupyter Express relies on a custom frontend extension that is activated when the Jupyter server starts, you must fully restart your Jupyter Notebook/Lab server after installing or upgrading Jupyter Express. Simply closing and reopening the browser tab is not enough.

---

Quick Start

Open a Jupyter Notebook and run the following commands in any cell:

import jupyter_express

%run_jupyter_express

A prompt widget will appear. Type your request, for example:

• "Plot a bar chart of sales by region"
• "Clean nulls from df and save to df_clean"
• "Train a random forest on X_train and y_train and print accuracy"

The agent then generates the code, and inserts it into a new cell above, ready to run.

---

Features

• Context-aware — before generating code, the agent analyzes the variables and dataframe columns present in your notebook, without you having to describe it. Note: Jupyter Express never sees your actual data; it only analyzes the names of your variables and dataframe columns.
• @ autocomplete — type @ to browse and insert variable or column names directly into your prompt
• Prompt history — press ↑ / ↓ in the widget to cycle through, remix and reuse previous prompts
• Persistent Agent Memory per notebook — remembers the last 15 prompts and responses within a notebook session, so follow-up prompts work naturally. The memory is resistant to kernel restarts and crashes, meaning you can pick up exactly where you left off each time.

---

Requirements

• Python 3.8+
• JupyterLab 4.x, Jupyter Notebook 7.x, or Jupyter Notebook 6.x (classic)
• Active internet connection (AI runs on a cloud backend)

---

Token Usage

On first use, a free trial is automatically activated whereby 25,000 tokens are assigned to each user without any sign-up. Tokens are consumed per request based on prompt and response length.

When your free tokens run out, a personalized link will be provided to you in the interface, which will take you straight to our website for top-up options

---

Changelog

1.1.0 — 2026-08-20

• Added support for Jupyter Notebook 7 and JupyterLab, in addition to Jupyter Notebook 6.

1.0.0 — 2026-06-26

• Initial release.

---

License

MIT License — see LICENSE for details.