Making simple Python wrapper kernels#

You can reuse IPython’s kernel machinery to easily make new kernels. This is useful for languages that have Python bindings, such as Hy (see Calysto Hy), or languages where the REPL can be controlled in a tty using pexpect, such as bash.

See also


A simple kernel for bash, written using this machinery

The Metakernel library makes it easier to write a wrapper kernel that includes a base set of line and cell magics. It also has a ProcessKernel subclass that makes it easy to write kernels that use pexpect. See Octave Kernel as an example.

If releasing a wrapper kernel as a Python package, see the steps in Packaging.

Required steps#

Subclass ipykernel.kernelbase.Kernel, and implement the following methods and attributes:

class MyKernel#

Information for Kernel info replies. ‘Implementation’ refers to the kernel (e.g. IPython), rather than the language (e.g. Python). The ‘banner’ is displayed to the user in console UIs before the first prompt. All of these values are strings.


Language information for Kernel info replies, in a dictionary. This should contain the key mimetype with the mimetype of code in the target language (e.g. 'text/x-python'), the name of the language being implemented (e.g. 'python'), and file_extension (e.g. '.py'). It may also contain keys codemirror_mode and pygments_lexer if they need to differ from language.

Other keys may be added to this later.

do_execute(code, silent, store_history=True, user_expressions=None, allow_stdin=False)#

Execute user code.

  • code (str) – The code to be executed.

  • silent (bool) – Whether to display output.

  • store_history (bool) – Whether to record this code in history and increase the execution count. If silent is True, this is implicitly False.

  • user_expressions (dict) – Mapping of names to expressions to evaluate after the code has run. You can ignore this if you need to.

  • allow_stdin (bool) – Whether the frontend can provide input on request (e.g. for Python’s raw_input()).

Your method should return a dict containing the fields described in Execution results. To display output, it can send messages using send_response(). If an error occurs during execution, an message of type error should be sent through send_response() in addition to an Execution results with an status of error. See Messaging in Jupyter for details of the different message types.

Kernel.send_response(stream, msg_or_type, content=None, ident=None, buffers=None, track=False, header=None, metadata=None, channel=None)#

Send a response to the message we’re currently processing.

This accepts all the parameters of jupyter_client.session.Session.send() except parent.

This relies on set_parent() having been called for the current message.

To launch your kernel, add this at the end of your module:

if __name__ == '__main__':
    from ipykernel.kernelapp import IPKernelApp

Now create a JSON kernel spec file and install it using jupyter kernelspec install </path/to/kernel>. Place your kernel module anywhere Python can import it (try current directory for testing). Finally, you can run your kernel using jupyter console --kernel <mykernelname>. Note that <mykernelname> in the below example is echo.


See also


A packaged, installable version of the condensed example below. will simply echo any input it’s given to stdout:

from ipykernel.kernelbase import Kernel

class EchoKernel(Kernel):
    implementation = 'Echo'
    implementation_version = '1.0'
    language = 'no-op'
    language_version = '0.1'
    language_info = {
        'name': 'Any text',
        'mimetype': 'text/plain',
        'file_extension': '.txt',
    banner = "Echo kernel - as useful as a parrot"

    def do_execute(self, code, silent, store_history=True, user_expressions=None,
        if not silent:
            stream_content = {'name': 'stdout', 'text': code}
            self.send_response(self.iopub_socket, 'stream', stream_content)

        return {'status': 'ok',
                # The base class increments the execution count
                'execution_count': self.execution_count,
                'payload': [],
                'user_expressions': {},

if __name__ == '__main__':
    from ipykernel.kernelapp import IPKernelApp

Here’s the Kernel spec kernel.json file for this:

{"argv":["python","-m","echokernel", "-f", "{connection_file}"],

Optional steps#

You can override a number of other methods to improve the functionality of your kernel. All of these methods should return a dictionary as described in the relevant section of the messaging spec.

class MyCustomKernel#
do_complete(code, cursor_pos)#

Code completion

  • code (str) – The code already present

  • cursor_pos (int) – The position in the code where completion is requested

See also

Completion messages

do_inspect(code, cursor_pos, detail_level=0)#

Object introspection

  • code (str) – The code

  • cursor_pos (int) – The position in the code where introspection is requested

  • detail_level (int) – 0 or 1 for more or less detail. In IPython, 1 gets the source code.

See also

Introspection messages

do_history(hist_access_type, output, raw, session=None, start=None, stop=None, n=None, pattern=None, unique=False)#

History access. Only the relevant parameters for the type of history request concerned will be passed, so your method definition must have defaults for all the arguments shown with defaults here.

See also

History messages


Is code entered in a console-like interface complete and ready to execute, or should a continuation prompt be shown?


code (str) – The code entered so far - possibly multiple lines

See also

Code completeness messages


Shutdown the kernel. You only need to handle your own clean up - the kernel machinery will take care of cleaning up its own things before stopping.


restart (bool) – Whether the kernel will be started again afterwards

See also

Kernel shutdown messages