papermill "Could not find kernel" in CI

papermill starts a kernel using the notebook's recorded kernelspec, or the one you pass with -k/--kernel. If that kernelspec is not registered on the runner, papermill cannot launch it and fails before executing cells.

What this error means

papermill fails with "jupyter_client.kernelspec.NoSuchKernel: No such kernel named X" or a message that it could not find the kernel, and jupyter kernelspec list lacks that name.

jupyter_client.kernelspec.NoSuchKernel: No such kernel named python3

Common causes

ipykernel not installed on the runner

No Python kernelspec is registered because ipykernel was never installed, so the requested kernel does not exist.

The notebook records a kernel name CI does not have

The saved kernelspec name (for example a conda env name) is not present in the CI environment.

How to fix it

Register the kernel and pass it explicitly

1. Install ipykernel into the active interpreter.
2. Register it as a named kernel.
3. Run papermill with -k set to that name.

python -m pip install ipykernel
python -m ipykernel install --user --name ci-kernel
papermill input.ipynb output.ipynb -k ci-kernel

List available kernels to confirm the name

Verify the kernel exists before running so the name in -k matches a registered kernelspec.

jupyter kernelspec list

How to prevent it

• Install ipykernel and register a known kernel name in CI.
• Pass -k to papermill instead of relying on the saved kernel name.
• Verify with jupyter kernelspec list in the setup step.

Related guides