deepelementlab/jupyter-studio

Jupyter Studio is an AI-powered extension for JupyterLab, the browser-based environment that data scientists and researchers use for writing and running code in interactive notebooks. The goal is to bring the kind of AI coding assistance that tools like Cursor offer to code editors directly into the notebook environment itself, without having to switch applications. Notebooks are organized into cells, individual chunks of code that you run one at a time and see results for immediately. Jupyter Studio understands this cell-based structure and adds AI features that work at that level. You can select code in a cell, press a keyboard shortcut, describe what you want changed in plain English, and see a highlighted diff to accept or reject. When your code produces an error, a single button click hands the problem to an AI agent that reads the error, looks at surrounding cells for context, edits the offending code, and re-runs it. There is also a chat panel that can reference specific cells or files by name, and a ghost-text completion feature that suggests code as you type, similar to GitHub Copilot. Beyond single-step help, the tool includes a multi-step agent mode: give it a higher-level instruction like refactor the data loader across cells 3 through 7, and it plans, edits, and runs cells in sequence, then reports what changed. You supply your own AI model credentials, choosing among Anthropic, OpenAI, Google, Azure, Ollama, vLLM, or any OpenAI-compatible endpoint, so your code stays on your machine unless you point it at a remote model. It ships as a JupyterLab extension, a pip-installable package, and a native desktop app for Windows, macOS, and Linux. It is free, open source, and licensed under Apache 2.0.