

agent:triage) to see what one agent writes.
Open a script to read its full source and its runs. From a run, the Script detail opens the same page.
Use it to see what your models actually write: which scripts run most, which fail, and which tools they reach for.