Skip to main content

​
What is Sphinx?

Sphinx is an AI assistant for data science that lives inside your code editor. It understands Jupyter notebooks, executes Python code, and helps you analyze data through natural conversation.

Installation

Get Sphinx installed in VS Code, Cursor, or Windsurf in minutes.

Getting Started

Follow our quickstart guide to run your first AI-powered analysis.

​
Supported Editors

Sphinx works with any VS Code-compatible editor:
EditorMarketplace
VS CodeVS Code Marketplace
CursorOpen VSX Registry
WindsurfOpen VSX Registry
Sphinx requires VS Code 1.96.0 or later (or equivalent editor version). The Microsoft Jupyter extension is automatically installed as a dependency.

​
Key Features

​
Notebook Integration

Sphinx deeply integrates with Jupyter notebooks. It can:
  • Create and edit cells — Add new code or markdown cells, modify existing ones
  • Execute code — Run cells and observe outputs, including DataFrames and visualizations
  • Understand context — Track variables, imports, installed packages, and the current working directory
  • Manage checkpoints — Automatically save notebook state for easy recovery

​
Operation Modes

Choose how much autonomy Sphinx has:

Agent Mode

Sphinx works autonomously, executing and iterating until the task is complete.

Plan Mode

Sphinx creates a plan first, letting you review before execution begins.

Safe Mode

Every code execution requires your explicit approval.

​
AI Capabilities

  • Web search — Query the web for documentation, tutorials, and current information
  • File search — Find, list, and read files in your workspace
  • Memories — Sphinx learns about your data and preferences over time
  • Rules — Configure custom instructions to guide Sphinx’s behavior

​
Keyboard Shortcuts

ActionmacOSWindows/Linux
Open SphinxCmd+TCtrl+T
If the shortcut conflicts with another extension, you can open Sphinx via the Command Palette: Sphinx: Open Chat

​
Commands

Access these commands through the Command Palette (Cmd+Shift+P / Ctrl+Shift+P):
CommandDescription
Sphinx: Open ChatOpen the Sphinx chat panel
Sphinx: New ConversationStart a fresh conversation
Sphinx: Open Sphinx RulesConfigure custom rules for Sphinx
Sphinx: Create Streamlit AppConvert your notebook into a Streamlit dashboard
Sphinx: Collapse Exploratory CellsHide intermediate exploration cells
Sphinx: RestartRestart the Sphinx extension
Sphinx: Log OutSign out of your Sphinx account
Sphinx: Open Welcome TourReplay the onboarding walkthrough

​
Settings

Customize Sphinx behavior in your editor settings (Cmd+, / Ctrl+,):
SettingDefaultDescription
sphinx.autosavetrueAutomatically save the notebook after each step
sphinx.enable-checkpointstrueKeep notebook checkpoints for recovery
sphinx.collapse-exploratory-cellsfalseAuto-collapse exploratory cells when generated
SettingDefaultDescription
sphinx.memory-readtrueUse learned insights to inform actions
sphinx.memory-writetrueLearn new insights about your data
sphinx.web-searchtrueAllow web searches for information
sphinx.require-web-search-approvalfalseRequire approval before each web search
SettingDefaultDescription
sphinx.file-searchtrueSearch, list, and read workspace files
sphinx.package-installationtrueAllow Sphinx to install new packages
sphinx.ripgrep-installationtrueInstall ripgrep for faster file search
SettingDefaultDescription
sphinx.enable-runtime-interruptstrueInterrupt cells that run too long
sphinx.show-edit-cell-wandtrueShow the edit wand icon in cells

​
Architecture

Sphinx consists of three components working together: 
┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│   VS Code       │     │   Sphinx        │     │   Jupyter       │
│   Extension     │◄───►│   Server        │◄───►│   Kernel        │
│                 │     │                 │     │                 │
│  - Chat UI      │     │  - AI Engine    │     │  - Python       │
│  - File access  │     │  - Tool calling │     │  - Execution    │
│  - State sync   │     │  - Planning     │     │  - Variables    │
└─────────────────┘     └─────────────────┘     └─────────────────┘
The extension communicates with Sphinx’s cloud server, which orchestrates AI reasoning and tool usage. The server instructs the extension to interact with the Jupyter kernel running in your local environment.
Your code runs locally in your Jupyter kernel. Sphinx sends code snippets to the server for AI analysis, but execution happens on your machine.

​
Next Steps

Install Sphinx

Step-by-step installation for VS Code, Cursor, and other editors.

Configure Rules

Set up custom rules to guide how Sphinx works with your projects.

Connect Data Sources

Link Snowflake, Databricks, and other data warehouses.

Explore Capabilities

Learn about the Sphinx Python library and AI capabilities.