An agent usually starts in a personal workspace, because that’s where someone had the idea. It shouldn’t stay there.
Move it to the team that owns the work
move transfers ownership. The original workspace no longer holds it.
Do this early. An agent in a personal workspace can’t be shared, and it leaves when the person does.
To keep a copy where it is:
copy is right when two teams want to evolve the same starting point in different directions. move is right when the agent has one owner and it isn’t you.
Make a variant
Duplicating inside the same workspace gives you a starting point for a variant — a version for a different region, or a different set of policies — without disturbing the original.
Make it discoverable
ORG_VISIBLE is what makes a good capability spread. Another team finds it, installs their own copy, and runs it with their own credentials. The original stays owned where it was built.
Make an agent org-visible once it has earned production — not while it’s still a draft. A discoverable agent that doesn’t work costs the next team more time than it saves.
Find and install someone else’s
Installing gives you your own installation in your active workspace. It runs with your credentials and appears in your installations:
From there you can schedule it:
To remove it:
Uninstalling removes your installation. The original agent is untouched.
Sharing with one person
Visibility is for capabilities the organization should reuse. To show one person one agent or one run, share it by email instead — see Share agents and runs.
Deleting
Deleting removes the agent and its history. Before deleting, consider setting it to PRIVATE, or moving it to an archive workspace — the run history is often the most valuable part.
See also
