**Claude Code Agent View** is a 📝Claude Code screen, opened with claude agents, for dispatching background sessions on your own machine and seeing at a glance which ones are working, done, or waiting on you.
Each background session is a full Claude Code conversation that keeps running with no terminal attached. Sessions are dispatched by typing a prompt into agent view, by running claude --bg "<task>" from a shell, or by sending a running session to the background with /bg. Rows show one of six states: Working, Needs input, Idle, Completed, Failed, or Stopped. Space peeks at a session's latest output or pending question, Enter attaches to it, and the left arrow on an empty prompt detaches; detaching never stops a session.
A session dispatched from agent view moves into its own git 📝Worktree under .claude/worktrees/ before it edits files, so parallel sessions do not collide. A background supervisor runs the sessions and stops ones left idle and unattached for about an hour, keeping their conversations on disk; pinning a row with Ctrl+T keeps it running. Shell commands such as claude attach, claude logs, claude stop, and claude agents --json manage sessions without opening the screen.
Agent view is a research preview, and its shortcuts may change. It is the local counterpart to 📝Claude Code Projects: agent view keeps the user as dispatcher for sessions that need the local machine, while a project hands dispatch to a coordinator running threads in the cloud. Agent view is distinct from the /agents slash command, which concerns subagent definition files.
Sessions dispatched from agent view can pass findings to each other through 📝Claude Code Cross-Session Messaging; 📝Claude Code Multi-Agent Orchestration compares every way to run Claude in parallel.