skop-runner 0.1.7

Local execution client for Skop Labs — run Jupyter notebooks from the browser on your own machine

Skop Runner

Run Jupyter notebooks from Skop on your own machine. Skop Runner connects your local computer to the Skop workspace, letting you execute code, manage files, and use your local Python environments — all from the browser.

Installation

Requires Python 3.10+.

# Recommended (isolated install, no venv needed)
uv tool install skop-runner

# Alternative
pipx install skop-runner

# Inside a virtual environment
pip install skop-runner

uv tool and pipx install skop-runner in its own isolated environment and add the CLI to your PATH. This is the recommended approach since skop-runner is a system-wide tool, not a per-project dependency.

Bare pip install into system Python is blocked on modern macOS/Linux (PEP 668). Use pip only inside an activated virtual environment.

Quick Start

1. Get your device token

Go to skoplabs.com, open a workspace, and click Add Device in the device dropdown. Copy the token.

2. Login

skop-runner login --token <your-device-token>

3. Start the runner

skop-runner start

That's it. Open your workspace in the browser and select your device — you can now browse folders, open notebooks, and execute code on your machine.

Commands

Command	Description
skop-runner login --token <token>	Authenticate with Skop
skop-runner start	Start the runner
skop-runner status	Check authentication status
skop-runner logout	Remove authentication

How It Works

Skop Runner runs a local process that:

• Connects to the Skop relay server via WebSocket
• Manages Jupyter kernels for notebook execution
• Serves your local files to the browser workspace
• Detects Python environments (venv, conda, Poetry, Pipenv, pyenv)

Your code runs entirely on your machine. Skop only relays messages between your browser and the runner.

Configuration

Config is stored in ~/.skop/config.json (created by skop-runner login).

Environment variable overrides:

• SKOP_API_URL — API URL (default: https://skoplabs.com)
• SKOP_WS_URL — WebSocket relay URL
• SKOP_LOG_LEVEL — Logging level (default: INFO)

Requirements

• Python 3.10+
• macOS or Linux (Windows support coming soon)

License

MIT