Claude Code Agent View Dashboards Multiple Agents
Simon Scrapesgo watch the original →
the gist
Claude Code's new terminal UI (v2.1.139+) manages multiple autonomous agents via a summary dashboard with status/repo sorting, session migration via /bg, reordering, pinning, and inline replies.
The Breakthrough
Claude Code ships Agent View, a native terminal UI that consolidates multiple agent sessions into a sortable, reorderable dashboard for monitoring status, jumping into details, and issuing quick replies.
What Actually Worked
- Developers update Claude Code to version
2.1.139or higher with theclaude updatecommand in the terminal. - Users launch the Agent View dashboard using the
claude agentscommand, which displays sessions organized by status. - Existing sessions migrate to the dashboard when developers run
/bgin the session terminal to background the task, freeing the terminal while preserving the session. - Sessions sort by status by default or by repository with
Ctrl+S; developers reorder them usingShift+UporShift+Down, and pin priority ones withCtrl+T. - New sessions start by entering a task description at the dashboard bottom, such as "draft a LinkedIn carousel based on my last YouTube video."
- Navigation uses right arrow to enter a session for full recap and reply, left arrow or backspace to exit; spacebar shows quick status previews across sessions for inline replies like approvals.
Context
Autonomous agents now handle tasks to 90% completion from good prompts, but developers previously juggled multiple terminals or used unofficial UIs like Vibe Kanban boards, tmux multiplexers, or custom command centers. Agent View provides Anthropic's native solution in research preview, limited to terminal (no desktop app yet) and lacking subfolder-level repo sorting. It pairs with an Agentic Operating System, a folder structure that injects client-specific context (e.g., clients/acme-corp for brand voice) into each session, enabling parallel work across repositories without terminal switching.
Notable Quotes
- "agents have gotten really good and we're no longer sitting in one chat GBT2 window babysitting every step and asking it to just redo everything they're now actually executing on a brief and getting us to 90% from a good prompt"
- "it's honestly about the time that Anthropic stepped in and shipped something native for us to all use"
- "the Aentic OS is just a folder structure at the end of the day that injects context at the right time"