Comments URL: https://news.ycombinator.com/item?id=48139837

Points: 3

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48139354

Points: 6

# Comments: 1

Comments URL: https://news.ycombinator.com/item?id=48138805

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48138370

Points: 1

# Comments: 1

Starting June 15, Pro plan subscribers can claim a $20 monthly credit for using the Claude Agent SDK and claude -p, including third-party tools built on the Agent SDK. As part of this change, Agent SDK and other programmatic usage will run on this credit, and will not impact your subscription limits. This includes third-party applications built on the Agent SDK. If you use your full Agent SDK credit in a given month, continued use will draw from extra usage, which can be manually enabled and disabled. Your subscription usage limits don’t change. They stay reserved for interactive usage of Claude Code, Claude Cowork, and chat. See the Help Center for more details.

Comments URL: https://news.ycombinator.com/item?id=48133373

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48131339

Points: 2

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48130679

Points: 203

# Comments: 44

Comments URL: https://news.ycombinator.com/item?id=48130256

Points: 4

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48127981

Points: 6

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48127794

Points: 2

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48127764

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48126691

Points: 2

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48126481

Points: 2

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48126438

Points: 49

# Comments: 36

Comments URL: https://news.ycombinator.com/item?id=48126429

Points: 9

# Comments: 8

This plugin implements the idea as Claude Skills, and adds a new ability: Automatic submit into Claude Code, without the need to manually copy and paste.

Installing the plugin has another benefit over "tell Claude to make HTML", since having it installed allows Claude to generate HTMLs on its own initiative, when it wants to show you something.

Comments URL: https://news.ycombinator.com/item?id=48124955

Points: 2

# Comments: 0

In the new world order of agentic development, poor performance is usually due to context bloat - either too many unnecessary tokens or missing relevant ones. I wrote this to help answer questions like:

- Which tool calls ballooned the context window, and on which turn? - What context was the subagent/skill passed when it was spawned? - What steps were repeated every session that could be lifted into a skill/documentation or cached lookup?

Agent Profiler is the tool for just that. It requires zero setup (just node), is fully local, open-source and can support any agent harness with a local transcript format.

Comments URL: https://news.ycombinator.com/item?id=48124878

Points: 1

# Comments: 0

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?

Comments URL: https://news.ycombinator.com/item?id=48124057

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48123922

Points: 2

# Comments: 0

cc-ledger is a local Rust binary that registers as a Claude Code hook, capturing runaway sessions and per-PR cost.

Cost analytics can be viewed from the macOS menu bar (SwiftBar plugin), a web dashboard at ccledger.dev, or the CLI.

All feedback and ideas appreciated.

Comments URL: https://news.ycombinator.com/item?id=48123808

Points: 3

# Comments: 0