Elyra ADE — documentation

Elyra ADE is an agent development environment — desktop and local: you build software by talking to and orchestrating several coding agents in parallel, with Canvas, Conversation, Terminal and Browser over the same project. Everything runs on your machine.

Where to start

If you want to…	Go to
see the product working from scratch	From scratch to your first project — add a disposable folder, open the first card, and delete everything at the end
look up a CLI command	Reference: elyra workspace list — flags, output, errors and limits
understand the mental model	Projects, screens and workspaces, below
organize work in cards	The Canvas
talk to an agent	The Conversation screen
use the project terminal	The Terminal screen
see your app running	The Browser screen
coordinate several agents	Orchestration: the conductor and the linked cards
work with git	Git in Elyra

Projects, screens and workspaces

Three words show up all the time and mean different things:

• A project is a repository or a working folder. It is the unit you open.
• A workspace is a name and a folder inside the project. The project’s screens open over the folder of the selected workspace. Several workspaces can point at the same folder, or at different ones.
• Screens are where you work: Canvas, Conversation, Terminal and Browser.

Two sentences that prevent most of the confusion:

1. Git follows the project folder; files follow the workspace folder. Git operations resolve the project’s repository; opening, reading and writing files resolves the workspace folder.
2. A workspace is not a git worktree. A worktree is a git feature and lives on the Git screen; elyra workspace list never lists git worktrees.

Technical names such as workspace, branch, commit, terminal and agents keep their original names — they are the names you find in the product and across the rest of the internet.