ajlab 0.1.3

Agent ready JupyterLab

ajlab

Agent-ready JupyterLab.

ajlab is a small meta-package for installing the JupyterLab pieces we use for agent-oriented workflows in one step. It does not add a new application layer of its own. Instead, it defines the dependency set and ships a little JupyterLab configuration under etc/jupyter/labconfig.

The project is aimed at providing a good agent experience using terminals directly within JupyterLab.

The goal is to keep the package easy to inspect: most behavior comes from the upstream Jupyter packages, while ajlab defines the default environment we want out of the box.

Approach

The package currently brings together:

• JupyterLab 4.6+
• collaboration and document-provider plumbing
• Jupyter server MCP support
• Git integration
• JupyterLab command-tooling support

These are installed as normal Python dependencies, so deployments can still override versions and Jupyter configuration in the usual ways.

Default settings

ajlab ships the following JupyterLab defaults:

• dockPanelPadding is disabled for the main shell, giving the workspace a tighter layout with less padding around docked content.
• Hidden files and folders are shown in the file browser by default. This is enabled both in JupyterLab's file browser setting and in Jupyter Server's contents manager so hidden paths are available to the browser.
• The extra human-collaboration UI surfaces are disabled by default: rtcPanel, shared-link, user-menu-bar, userEditorCursors, and userMenu.

The collaboration and document-provider packages are still installed. The default UI is just kept quieter for agent-centered sessions.

The JupyterLab frontend defaults are shipped as labconfig/*.d/00-ajlab.json fragments so downstream meta-packages can add their own Lab configuration fragments without replacing ajlab's files.

Install

pip install ajlab

License

BSD-3-Clause