termdeck
Agent view is the session roster built into Claude Code. Run claude agents, or press the left arrow on an empty prompt, and every background session on the machine shows up as a row: its state, its recent activity, and what it is waiting on. Type a prompt to dispatch a new background session. A per-user supervisor process keeps dispatched sessions running with no terminal attached, so you can close the shell and come back later. Peek into any row to read the tail, reply, or approve a permission request, and attach fully when one needs real attention. It launched as a research preview on 11 May 2026 and is included on Pro, Max, Team, Enterprise, and API plans.
Comparison
Termdeck vs Claude Code agent view: beyond one machine.
Agent view is Claude Code's built-in roster for background sessions on the machine in front of you; Termdeck is a browser board for every Claude, Codex, and Grok session on every machine you own.
Context
What Claude Code agent view is
Side by side
Claude Code agent view vs Termdeck, feature by feature
| Agent view | Termdeck | |
|---|---|---|
| Agent CLIs supported | Claude Code only | Claude Code, Codex, and Grok; engine picked per chat |
| Where you drive it | Terminal UI: claude agents | Browser SPA + installable PWA |
| Machines at once | One; sessions stop if the machine shuts down | Whole mesh, one owner-scoped board |
| Approve from your phone | No; terminal notifications while the roster is open | Web Push that deep-links into the session |
| Terminal resume still works | Yes; it is the CLI | Yes; disk is the source of truth |
| Diff review | Terminal output | Every edit rendered as a diff |
| Self-hostable | N/A; local software, nothing to host | No; hosted master, while agents and transcripts stay on your machines |
| Platforms | Wherever Claude Code runs | Agents on macOS, Linux, and Windows |
| Price | Included with Claude Code (Pro, Max, Team, Enterprise, API) | Free tier, then $9.99/mo |
Verified 19 August 2026 against the Claude Code docs and Anthropic's launch post. Agent view is a research preview and moves fast; tell us if a row is out of date.
Straight answer
Where agent view wins
If your work is Claude Code on the machine in front of you, agent view is the shortest path to running sessions in parallel: it ships inside the CLI you already have, costs nothing beyond the plan you already pay for, and needs no account, no installed agent, and no network path to anyone's server. Dispatch is one keystroke from the terminal you were already in. And because it is first-party, new Claude Code capabilities can land in agent view before any third-party surface can follow. On setup time, Termdeck does not beat it and does not try.
The delta
Where Termdeck wins
Every machine, one board
Agent view rosters the sessions on one machine, and they stop if that machine shuts down. Termdeck's agents dial out from every box you own, so laptop, desktop, and homelab sessions land on one owner-scoped board, with Web Push that reaches your phone when a run blocks.
Three engines, one surface
Agent view is Claude Code only. Termdeck gives Claude Code, OpenAI Codex, and xAI Grok one sidebar, one transcript, and one permission surface, with the engine picked per chat, so a mixed-engine day does not mean three windows.
Rate limits meet more accounts
Anthropic's own docs note that ten parallel background sessions burn plan quota roughly ten times as fast. Termdeck lets you connect more than one Claude account and switch mid-session when a limit hits, so the run keeps going instead of parking until the window resets.
Questions
Termdeck vs agent view, answered
What is Claude Code agent view?
Agent view is a session roster inside the Claude Code CLI, opened with claude agents. It lists every background session on the machine, lets you dispatch new ones, peek at their output, reply, and approve permission requests. Anthropic shipped it as a research preview on 11 May 2026, and it is still labelled one today.
How to view Claude Code agents?
On one machine, run claude agents and every background session appears as a row you can peek into or attach to. Seeing sessions across several machines, or from your phone, takes a separate surface: Termdeck shows every session on every connected machine in one browser board, and Anthropic's Remote Control can open a single session on another device.
Why aren't developers convinced by agent view?
The recurring gripes in early coverage and Reddit threads are about reach, not the idea: agent view is a research preview whose interface may still change, its sessions live on one machine and stop when that machine shuts down, and the roster is Claude Code only, in the terminal only. If you want the same list to follow you across machines, engines, and onto a phone, that is the part a separate control surface adds.
What is the best agentic coding tool?
That question is about the agent itself: Claude Code, Codex CLI, and Grok CLI are the coding tools, and which is best depends on your stack and your plan. Agent view and Termdeck sit one layer up. They do not write code; they let you see and steer the sessions those tools run. Termdeck stays neutral by driving all three engines from one board.
Keep reading
Related
Claude Code web UI alternatives
The wider field: seven ways to put Claude Code in a browser, compared honestly, including the ones that are not Termdeck.
Termdeck vs tmux
The terminal lane: what a multiplexer still does best, and where an agent board takes over.
How sessions work in Termdeck
Disk is the source of truth: how the browser and the terminal share one transcript without forking it.
Also see Claude Code web UI and Termdeck vs Claude Code Remote Control.