Skip to content

INSTALL.md

Latest commit

 

History

History
75 lines (54 loc) · 2.3 KB

INSTALL.md

File metadata and controls

75 lines (54 loc) · 2.3 KB
Raw

Install openLearn

openLearn supports Python 3.11 through 3.13 on Linux, macOS, and Windows. The command line and Maker Bench use the same local home and files. Docker or Podman is optional unless you run secure Python code checks.

Install

Create a virtual environment before installing when practical.

macOS and Linux

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip openlearn
openlearn --version

Windows PowerShell

py -3.11 -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip openlearn
openlearn --version

If PowerShell blocks activation, follow your organization's execution-policy guidance or invoke .venv\Scripts\python.exe -m pip directly. pipx install openlearn is also suitable when you want an isolated command-line installation managed by pipx.

First launch

openlearn

Bare openlearn opens Maker Bench in the default browser. Run openlearn cli to use the terminal interface instead. openlearn init remains available when you want to configure a provider directly from the terminal. You can use a hosted OpenAI-compatible provider with your own account and key, or a local keyless endpoint such as Ollama. openLearn does not supply, bill for, or share a provider account for Community users. Maker Bench is loopback-only and opens in the default browser. Use openlearn web --no-browser on a headless machine, or openlearn web --port 9000 to select a loopback port explicitly.

Upgrade and uninstall

Upgrade the package in the same environment where you installed it.

python -m pip install --upgrade openlearn
openlearn --version

To remove only the installed package:

python -m pip uninstall openlearn

Package upgrades and uninstalls do not remove learner data. Before changing machines or deleting data, create a verified backup as described in Data and privacy.

Optional code runner

The code workbench remains available without a container runtime. Secure execution needs Docker or Podman. If execution is unavailable, install or start Docker or Podman and run:

openlearn doctor

Do not treat --reduced-isolation as a secure sandbox. It runs learner code locally with warnings and can access account files and the network.