Spent half a day trying Claude Code 2.1.139’s new Agent View and background sessions — useful, but still has quite a few rough edges.

The first item in the 2.1.139 changelog released on 2026-05-11 was Added agent view (Research Preview). Interestingly, they even included a dedicated doc link in the changelog this time: https://code.claude.com/docs/en/agent-view

Not sure how many people here have tried it yet. I spent a few hours with it today and found both some really nice ideas and some frustrating issues. Curious to hear other people’s experiences too.

1. What is Agent View?

Agent View can launch parallel Claude sessions directly, and each session automatically uses a separate git worktree for isolation, so multiple sessions can run in parallel without conflicts.

The background sessions are managed by an independent supervisor process. They continue running even if the terminal or Agent View itself is closed. Closing the terminal does NOT stop the background sessions.

Checking processes with ps, you can see a supervisor process like:

claude daemon run --spawned-by xxxx

with PPID = 1.

I honestly never noticed this supervisor process before. I downgraded back to 2.1.138 just to test, and it seems older versions don’t start it. In 2.1.139, using claude agents, claude --bg, or /bg will spawn the supervisor.

2. My experience so far

After using it for a while today, a few usability issues stood out immediately:

- Sessions always create worktrees, but Agent View doesn’t show branch status (mergeable, already merged, etc.), so I constantly had to check branch state outside Agent View - Agent View shows ALL background sessions globally, with no workspace/project filtering. If you use background sessions across multiple repos, the list becomes messy very quickly - When dispatching sessions from Agent View, you can’t specify permission mode. You have to use claude --bg or /bg first to configure permissions, then “push” it into Agent View afterward

Honestly, this feature feels a bit rushed compared to some earlier Claude Code features that were hidden behind explicit environment flags. Agent View is the opposite — it’s enabled by default, and you need an environment variable to disable it.

That said, after using it overall, I do think it can partially replace tools like Conductor.

Personally, I feel the most important change in this release is actually the introduction of the supervisor role itself. It opens the door for much more functionality in the future, especially around remote-control capabilities.

I also tested this with remote-control enabled for all sessions. Background sessions created from Agent View are visible inside the Claude mobile app, and you can continue chatting with them there, which is honestly pretty cool.

Right now the Claude mobile app still can’t create sessions directly, but now that Claude Code already has a supervisor architecture in place, I wouldn’t be surprised if starting sessions on your host machine from the mobile app becomes possible very soon. Remote-control feels like it’s going to get much more powerful over time.

And finally, a small shameless plug

If you can’t use remote-control, or just don’t like the chat-style interaction model, you can try the remote Claude Code / Codex app I built: ServerCC https://servercc.llpsh.com

It’s also searchable on the App Store.

Besides enabling high-quality remote Claude Code/Codex usage over mobile networks, it also supports:

browsing and uploading workspace files

VNC desktop viewing for UI debugging

Termius-style persistent sessions / keep-alive

and more

What are your thoughts on features like Claude Code Agent View or background agents?

Burn – K8s cost waste by namespace and pod. Just kubectl, no deploy