npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

orion-notebook

v0.14.0

Published

Local Orion CLI launcher for notebooks and Jupyter

Downloads

2,015

Readme

Orion is the first open-source AI coding agent built specifically for data scientists, analysts, and researchers.

General-purpose coding agents weren't designed for data-heavy research workflows. They guess at column names, choke on large datasets, and optimize for code that runs rather than results that make sense. The root cause is the context gap between what a model can read from source files and what it can't see: your runtime state, live variable values, cell outputs. The actual data you're working with.

Orion closes that gap. Built around a modern notebook IDE, it feeds the right context to the model at the right time: variable inspector state, cell execution order, rich outputs, tracebacks, and dataframe structure. The agent iterates alongside you like a research partner — observing inputs, checking results, and adjusting until the outcome actually looks right.

Demo

Make focused edits with confidence, or generate entire notebooks from scratch.

Use any model, from any provider. Choose remote or local inference.
Mention cells, variables, files, etc. so you and the agent are always on the same page.

Invoke commands, skills and sub-agents with slash commands.

Quick start

Prerequisites: The install script bootstraps Orion with a managed uv tool environment by default. Orion can create its own Python/Jupyter runtime for notebooks on first run.

Running orion starts Jupyter, launches Orion locally, opens your browser, and auto-connects to the Jupyter server. All managed files live under ~/.orion (Windows: %USERPROFILE%\.orion).

Install script

macOS / Linux:

curl -fsSL https://www.orion-agent.ai/install.sh | bash

Windows (PowerShell):

powershell -ExecutionPolicy Bypass -c "iwr -useb https://www.orion-agent.ai/install.ps1 | iex"

The script installs orion-notebook with uv-managed Python 3.12 so the orion command is available across shells. Pin a version with ORION_VERSION=0.9.0 or force an advanced method with ORION_INSTALL_METHOD=npm|pip|uv.

Install via npm (advanced)

Both npm and PyPI use the package name orion-notebook:

npm install -g orion-notebook
orion

Or run once without installing globally:

npx orion-notebook

The npm package ships the full Orion app bundle, so no separate app download is needed. You only need Node.js 20+ installed; Python 3.8+ is required for notebook execution.

Use npm when you want Orion to use an existing Python or conda environment and remember that choice:

python -m pip install -U jupyter_server ipykernel orion-ui
orion config python pick
orion

First run: if Orion does not find a compatible Jupyter setup, it prompts to create an Orion-managed environment under ~/.orion/runtime/venv. That step installs Jupyter into Orion's venv (not your global Python) and can take a few minutes. Approve automatically with:

orion --yes

Install via pip

python -m pip install orion-notebook
orion

Use pip when you want Orion installed inside one specific Python environment. If that same environment should run notebooks for Orion:

python -m pip install -U orion-notebook jupyter_server ipykernel orion-ui
orion

The PyPI wheel is small by design. On first run it may:

  1. download the Orion app bundle into ~/.orion/app/<version>
  2. download portable Node.js 20+ into ~/.orion/runtime/node if Node is missing
  3. create an Orion-managed Jupyter venv under ~/.orion/runtime/venv if needed (includes orion-ui for notebook UI components — no extra install step)

Each step prompts before downloading or installing unless you pass --yes:

orion --yes

After the first successful setup, later runs are much faster.

Diagnose an install

If setup fails or behaves differently across machines, run:

orion doctor --json
orion doctor --setup --json

doctor --setup performs first-run setup checks without opening a browser, which is useful for CI and support reports.

Update Orion

Orion checks for stable updates before starting and again while the app is open. To update explicitly through the package manager that installed Orion, run:

orion update

After a CLI update, run orion again. Desktop updates download in the app and ask before restarting.

Adding yourAPI Keys

Open Orion in your browser, then go to Settings → Providers to add your API keys.

For manual Jupyter setup instead of the CLI, connect a Jupyter server in the GUI (supports Jupyter Server 1.x and 2.x when required APIs are available).

Run from source (developers)

Copy .env.example to .env in the repo root before building or running the dev server. Next.js loads .env, not .env.example.

macOS / Linux:

git clone https://github.com/nicolasakf/Orion-app.git
cd Orion-app
cp .env.example .env
npm install

Windows (PowerShell):

git clone https://github.com/nicolasakf/Orion-app.git
cd Orion-app
Copy-Item .env.example .env
npm install

Dev mode

Both commands start the Next.js dev server on port 3001 with hot reload (Turbopack). Pick one:

| Command | What it starts | When to use | | --- | --- | --- | | npm run dev | Next.js only | UI and app work. Connect Jupyter yourself in the app (kernel selector or an existing server). | | npm run dev:notebook | Jupyter + Next.js (same bootstrap as the orion CLI) | Notebook and kernel work. Orion auto-connects via ~/.orion/runtime/jupyter-connection.json. |

npm run dev              # Next.js dev server only
npm run dev:notebook     # Jupyter + Next.js (recommended for notebooks)

Useful dev:notebook flags: --here (Jupyter root = current directory), --pick-python (interactive Python selection). Setup prompts are auto-approved by default; set ORION_DEV_YES=0 to require interactive consent.

Production-like CLI run

To test the packaged orion CLI locally (not the dev server):

npm run build
npm run build:cli
npm run prepare:app-bundle
node dist/cli/cli/orion.js

See Contributing for more development details and CLI development for flags, publishing, and local CLI testing.

Uninstall

pip uninstall and npm uninstall remove the launcher only. To also remove cached Orion data (especially the pip-downloaded app bundle under ~/.orion/app/<version>/), run:

orion uninstall --yes

Remove everything under ~/.orion (Jupyter venv, portable Node, all cached app versions):

orion uninstall --all --yes

Then remove the package:

pip uninstall orion-notebook
# or
npm uninstall -g orion-notebook
# or
uv tool uninstall orion-notebook

Links

| Description | Link | | --- | --- | | Website | orion-agent.ai | | Hosted app | app.orion-agent.ai | | User docs | docs.orion-agent.ai |

Documentation

  • User help — install guides and troubleshooting (Orion-docs)
  • Contributing — setup, CLI development, publishing, pull requests
  • PyPI package — pip install details and first-run behavior
  • Architecture — how the app fits together
  • Agent API — tools, skills, sub-agents, adding models

Tech stack

Next.js 15, React 19, Tailwind CSS, Monaco Editor, Vercel AI SDK, JupyterLab services.

License

Apache-2.0. See LICENSE.