jupyter-kernel-client

What it is and what it does

Jupyter Kernel Client is a Python library that lets you programmatically connect to and execute code in Jupyter Kernels running on a remote Jupyter Server. It communicates via HTTP and WebSocket protocols to send code to a kernel and retrieve execution results, outputs, and status. The package wraps the kernel protocol in a simple context-manager API and also provides an optional interactive console (Konsole) for REPL-style interaction.

Typical use cases include automating code execution in a kernel from an external script, building tools that need to run code in a Jupyter environment without launching a notebook interface, or integrating Jupyter computation into larger applications. The package depends on standard Jupyter libraries (jupyter-client, jupyter-core, traitlets) and network utilities (requests, websocket-client) to handle the protocol details.

Use it for:

• Automate code execution in a remote Jupyter kernel from a Python script or application
• Build a custom REPL or interactive console that connects to an existing Jupyter kernel
• Integrate Jupyter computation into a larger application or service without a notebook UI
• Execute code in a kernel and capture structured results (outputs, execution count, status)
• Connect to an existing kernel in JupyterLab and run code programmatically

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Connects to live Jupyter Kernels over HTTP and WebSocket to execute code remotely, and provides an interactive console (REPL) for kernel interaction.

Yes. The package is actively maintained, has low install friction, carries a permissive BSD license, and solves a clear problem—remote kernel execution—for developers building Jupyter-integrated tools. No known vulnerabilities. Install if you need to execute code in a Jupyter kernel from outside a notebook.

Install

jupyter-kernel-client on PyPI

pip install jupyter-kernel-client

uv add jupyter-kernel-client

poetry add jupyter-kernel-client

Installing jupyter-kernel-client

Before you install

Low friction install with 7 runtime dependencies, all standard Jupyter ecosystem packages. Actively maintained with a release 6 days ago and recent commits.

License in practice

BSD 3-Clause License permits commercial and private use with attribution and liability disclaimer; no restrictions on distribution or modification.

Quickstart

pip install jupyter-kernel-client

from jupyter_kernel_client import JupyterKernelClient

with JupyterKernelClient(server_url="http://localhost:8888", token="MY_TOKEN") as kernel:
    reply = kernel.execute("print('hello')")
    print(reply)

Requires a running Jupyter Server with ipykernel; Python >=3.10

Verify before relying

• Performance characteristics and latency when executing code over HTTP/WebSocket
• Support for kernel features like interrupt, restart, and shutdown operations
• Compatibility with different Jupyter Server versions beyond current release

Package facts

Evidence: jupyter_kernel_client-1.0.1-py3-none-any.whl

Keywords: Jupyter

Tags

More Application Frameworks packages