Claude Code /agents Gets a Tabbed Layout: Running and Library Tabs for Subagent Management

A Tabbed Interface for Multi-Agent Management

Claude Code v2.1.98 redesigns the /agents command with a tabbed layout, splitting the agent management view into two distinct, purpose-built panels: a Running tab and a Library tab. The change reflects the growing complexity of agentic workflows in Claude Code — as Dispatch, Channels, and Managed Agents have made multi-agent sessions increasingly common, the flat /agents view was no longer sufficient to manage them clearly.

The Running Tab

The Running tab provides a live view of all currently executing subagent instances within the session. Developers can see which agents are active, inspect their status, and take contextual actions on running instances — such as viewing their current task, interacting with their output, or terminating them. This gives a clear operational picture of what is happening in the background at any given moment, which is especially important in complex sessions where several subagents may be executing in parallel.

Previously, tracking live subagents required mentally correlating Dispatch calls and Channels messages. The Running tab centralizes this into a single, readable view.

The Library Tab

The Library tab serves as the catalog of available agents — the place to browse, configure, and deploy new subagents. From this tab, developers can:

• Browse saved agent definitions and templates
• Launch a new instance of an agent directly into the session ("Run agent" action)
• Navigate to a running instance of an already-deployed agent ("View running instance" action)

The Library/Running separation mirrors a model familiar to developers from process managers and service orchestrators: there is a registry of what can run, and a separate operational view of what is running. Applying that pattern to Claude Code's agent system makes the interface substantially more predictable.

Why This Matters for Agentic Workflows

Claude Code's investment in multi-agent infrastructure has accelerated significantly in recent months. Dispatch (for launching subagents from within a session), Channels (for structured inter-agent communication), and Managed Agents (persistent agent definitions that can be reused across sessions) have collectively transformed Claude Code from a single-agent coding assistant into a platform for orchestrating multiple specialized agents working in concert.

As these capabilities have grown, the UX for managing agents has lagged behind. The tabbed /agents layout is a direct response to that gap — it provides the control surface that multi-agent workflows now require.

Context: Recent Agent UX Improvements

This release is not an isolated change. Claude Code has been iterating on agent management UX across several consecutive versions:

• v2.1.88 introduced foundational improvements to the /agents command interface
• v2.1.89 added the ability to view running subagent instances from within /agents
• v2.1.97 refined agent lifecycle controls and status visibility
• v2.1.98 (this release) restructures the entire interface into a two-tab layout with Running and Library views

Taken together, these releases reflect a deliberate, incremental effort to build a production-grade agent management experience inside Claude Code's terminal UI. The tabbed layout is the most significant UX consolidation in this sequence — it brings together the capabilities added across previous versions into a coherent, unified interface.