nteract

Spec sheet

Product

nteract describes itself as a local-first notebook environment where people, kernels, and AI agents can work with the same live document.Read 5 Oct 2026 · github.com

Agent integrations

The official agent setup describes plugins for Claude Code and Codex that provide notebook-backed Python sessions with persistent variables and rich outputs.Read 5 Oct 2026 · nteract.io

Data exploration

The product offers an interactive DataFrame explorer with crossfiltering for Python data including pandas, polars, datasets, narwhals, and duckdb.Read 5 Oct 2026 · nteract.io

Desktop app

The desktop app bundles the notebook, the runt CLI, and the runtimed daemon, and is available for macOS, Windows, and Linux.Read 5 Oct 2026 · nteract.io

Headless installation

A per-user headless installer is offered for Linux and macOS and sets up the app, CLI, daemon, and a user service without root access.Read 5 Oct 2026 · nteract.io

Local data

The security page says notebooks, outputs, and environments live on the user's filesystem and that no account is required.Read 4 Oct 2026 · nteract.io

Maintainer

The telemetry page says nteract is maintained under NumFOCUS, a nonprofit, and is a NumFOCUS sponsored project.Read 4 Oct 2026 · nteract.io

Markdown and TeX

nteract 2.6 added improved Markdown rendering and renders bare TeX environments as math.Read 5 Oct 2026 · nteract.io

MCP

The nteract MCP server lets AI assistants run code, read and write cells, manage dependencies, and work with people on the same live notebook.Read 5 Oct 2026 · github.com

MCP and CLI

The nteract CLI and MCP server let agents create and manage notebooks, run code, read outputs, and manage dependencies.Read 4 Oct 2026 · github.com

Notable limit

The installer page documents stable and nightly channels, with nightly installed side-by-side using channel-suffixed commands.Read 5 Oct 2026 · nteract.io

Notebook dependencies

Users can store dependencies in notebook metadata, and nteract requires approval before installing them.Read 4 Oct 2026 · nteract.io

Notebook editing

Users can navigate long notebooks with a live outline and manage packages from a left rail.Read 5 Oct 2026 · nteract.io

Notebook workflows

The desktop app opens Jupyter notebooks and includes the notebook, the runt CLI, and the runtimed daemon.Read 4 Oct 2026 · nteract.io

Output isolation

The security page says notebook outputs render in an opaque-origin iframe that cannot access the parent app's DOM, localStorage, or Tauri APIs.Read 4 Oct 2026 · nteract.io

Package safety

nteract says it checks for typosquatting in popular PyPI packages and warns users about suspected package typos.Read 4 Oct 2026 · nteract.io

Runtime environments

nteract detects project environments from pyproject.toml, pixi.toml, and environment.yml, and supports uv, conda, and pixi.Read 4 Oct 2026 · nteract.io

Secret redaction

nteract can scrub secret-looking environment values from outputs before they are written to disk or shown to an agent, and this feature can be disabled in settings.Read 5 Oct 2026 · nteract.io

Telemetry

The maker says telemetry is an anonymous daily heartbeat with six build-related fields, can be disabled, and does not include notebook contents or cell outputs.Read 5 Oct 2026 · nteract.io

Telemetry retention

The telemetry page says raw pings are retained for 400 days and daily aggregate counts are retained indefinitely.Read 5 Oct 2026 · nteract.io

Tracebacks

Python errors can render as structured, navigable tracebacks that link to the cell that raised the error.Read 5 Oct 2026 · nteract.io

Where Specifiction places it

• Notebook Software#22 of 257.2