Skip to main content

​
Overview

Sphinx is built for Jupyter notebooks. It reads your notebook’s cells, understands your kernel’s state, and generates code that integrates with your existing work.
Sphinx requires a Python kernel. It automatically validates your kernel selection and prompts you to select or install one if needed.

​
Key Capabilities

Create & Run Notebooks

Learn how Sphinx adds cells, edits code, executes cells, and understands your notebook’s current state.

Connect to Backends

Use Sphinx with Databricks, Google Colab, Azure ML, JupyterHub, and other notebook environments.

Long-Running Operations

How Sphinx handles time-consuming tasks like model training and large data processing.

​
How Sphinx Interacts with Notebooks

When you chat with Sphinx, it:
  1. Reads your notebook — Sees all cells, their code, and outputs
  2. Extracts kernel state — Knows your variables, dataframes, imports, and installed packages
  3. Generates contextual code — Creates code that uses your existing variables and follows your patterns
  4. Executes and observes — Runs cells and learns from the results
  5. Iterates as needed — Fixes errors, refines visualizations, and continues until the task is complete

​
What Sphinx Knows About Your Notebook

ContextDetails
CellsAll code and markdown cells, including outputs
DataFramesNames, shapes, column types, and sample rows
VariablesVariable names and types (up to 300)
ImportsModules you’ve imported
PackagesInstalled packages with versions
EnvironmentAvailable environment variable names
This context allows Sphinx to write code that integrates with your existing work. Using your variable names, respecting your data structures, and building on what you’ve already done.

​
Supported Environments

Sphinx works anywhere VS Code’s Jupyter extension works:
  • Local Python — Conda, venv, system Python
  • Cloud Notebooks — Databricks, Google Colab, Azure ML, SageMaker
  • Remote Servers — JupyterHub, remote Jupyter servers
  • Containers — Docker, Dev Containers, Codespaces
See Connect to Different Backends for setup guides.

​
Quick Start

1

Open a notebook

Open any .ipynb file in VS Code, or let Sphinx create one for you.
2

Start the kernel

Select a Python kernel when prompted, or run a cell to activate it.
3

Open Sphinx

Press Cmd+T (macOS) or Ctrl+T (Windows/Linux) to open the Sphinx panel.
4

Start chatting

Describe what you want to do. Sphinx will add, edit, and execute cells in your notebook.

​
Operation Modes

Sphinx offers three modes to match your workflow:
ModeDescriptionBest For
SafeEvery execution requires approvalLearning, sensitive environments
AgentAutonomous with guardrail checksEveryday workflows
PlanCreates a plan before executingComplex, multi-step tasks
Switch modes using the dropdown in the Sphinx chat panel.