See it work
Describe an agent. See it hosted.
Give it a name, a job and some tools. You see how Agent host would hold it: what it may do alone, what waits for a person, and a sample run.
A simulation, in your browser. The definition, the run log and the inbox are filled from the built-in templates with your words. Nothing is created, no tool runs and nothing is deployed. The way the risk rule treats each tool is the real rule; the run itself is a scripted sample with sample data. Nothing is uploaded, unless you switch on the small assistant below.