Claude Code Agent View: How Anthropic’s New CLI Dashboard Manages Parallel AI Coding Sessions

The official Agent View documentation says it is opened with:

claude agents

From that view, developers can dispatch new sessions, watch their state at a glance, and step into a session only when it needs attention . The docs also state that sessions can keep running in the background without a terminal attached, which is the core workflow shift: Claude Code sessions become background work items that can be supervised from one place .

Why it matters for developers

Agent View is most useful when Claude can work on several independent tasks at once. Anthropic’s documentation gives examples such as fixing a bug, reviewing a pull request, or investigating a log . In those cases, the developer’s job is less about typing continuously into one chat and more about coordinating parallel work.

That coordination is exactly what Agent View targets. Without a shared dashboard, developers may need to switch between terminal tabs, panes, or transcripts to find out which session is blocked and which is still making progress. Agent View centralizes that status information so the developer can intervene only when a session needs input . Testing Catalog similarly describes the feature as a command-line control center for parallel Claude Code work, including launching agents, sending sessions to the background, checking running tasks, and jumping back into full sessions when needed .

The key workflow: claude agents

For users on Claude Code 2.1.139 or later, the basic flow is simple:

claude --version
claude agents

The first command checks the installed Claude Code version, and the second opens Agent View . Once inside, the dashboard shows the consolidated session list rather than requiring the developer to inspect each session separately .

Agent View surfaces three practical states:

That makes Agent View less like a new model feature and more like operational tooling for managing multiple AI coding processes.

How /goal fits into Claude Code 2.1.139

Agent View was not the only developer-facing addition in Claude Code 2.1.139. The same official changelog also added /goal, a command that lets the developer set a completion condition so Claude keeps working across turns until that condition is met .

The changelog says /goal works in interactive mode, -p, and Remote Control, and shows live elapsed time, turn count, and token usage in an overlay panel . In practice, the two features complement each other: /goal gives a session a defined stopping condition, while Agent View gives the developer a way to monitor multiple sessions from one command-line dashboard.

Who can access Agent View

The official availability requirement in the provided sources is Claude Code version 2.1.139 or later. Users can verify their installed version with claude --version and start the feature with claude agents .

Some third-party coverage reports plan-level availability on Pro, Max, Team, Enterprise, and Claude API plans . The official changelog excerpt provided here, however, frames access around the Claude Code version rather than plan eligibility, so teams should confirm access in their own Claude Code environment before standardizing workflows around it .

Bottom line

Agent View turns Claude Code into a more manageable tool for parallel AI coding work. It does not change the basic need for developers to supervise, respond, and review; it changes where that supervision happens. Instead of watching scattered sessions one by one, developers on Claude Code 2.1.139 or later can use claude agents to see what is running, what is waiting on them, and what has finished from a single CLI dashboard .