Subagents

When Claude Code starts a subagent, through the Agent tool or a forked skill, the subagent writes a transcript of its own. It appears in snoop under the call that started it.

Expand inline

Press x on an Agent call to expand what its subagent did, indented under a │ rule. The expanded events stay live while the subagent runs. Press X to expand every subagent, nested ones included, and again to collapse them all.

Open in its own view

Press a on an Agent call to open the subagent’s transcript in a view of its own, with all four tabs. Press esc to return to the parent session.

Search and files

Search and filters reach inside subagents. A collapsed call whose subagent has matches stays in the list, marked ●. The Files tab includes files the subagents touched, labeled with the subagent that touched them.

Waiting and stalls

A subagent can run for minutes, so snoop never reports an Agent call, or a forked skill whose subagent is writing, as waiting on you. See Alerts and notifications.