That is cool but this seems to be only for simulating user behavior. Form the docs: `Test real agent behavior by simulating users in different scenarios and edg

That is cool but this seems to be only for simulating user behavior. Form the docs: `Test real agent behavior by simulating users in different scenarios and edge cases`. What I mean by simulation is more on simulating a situation an agent may come accross in production. All read-only tool calls return simulated responses based on the environment state. Write actions change the environment according to scenario.