Google’s overview describes a desktop command center for launching and inspecting agents, a terminal client, editor extensions and a Python SDK. It also documents visual artifacts, subagents, MCP and permissions. The exact capabilities differ by surface, so a comparison should name the app, CLI or extension being evaluated.
For example, a team interested in orchestration should inspect plans, diffs and task isolation. A developer who wants to keep an existing editor should inspect the supported extension. Those are different needs, even when the same product name and model appear in the interface.