In the next version of Claude Code: subagents run in the background by default, so you can keep talking to Claude while your subagents work
If you want your agent to run in the foreground, just tell Claude
10:31 AM · Jun 29, 2026 · 122.6K Views
Claude Code creator Boris Cherny says the next version will run subagents in the background by default
Story Overview
Boris Cherny, who created and leads Claude Code at Anthropic, announced that the next release will flip subagents to background mode by default so the main conversation can keep rolling while independent tasks finish on their own.
2412.4K92244102.5K
Developer Impact
Background agents free up your main thread
Instead of pausing to watch every subagent step, users can stay in the primary chat and simply flag when they want something pulled back into the foreground.
Open Question
Exact rollout timing stays unspecified
Cherny’s note gives no date, access tier details, or phased rollout plan, leaving the practical arrival of the change open for now.
Sentiment
Many users welcome Claude Code running subagents in the background by default as a huge UX and workflow win for multitasking, while some criticize it for wasting tokens or voice unrelated complaints about Anthropic.
Pos
81.7%
Neg
18.3%
84 comments with sentiment.
Cluster Engagement
Digg Deeper
No Digg Deeper questions have been answered for this story yet.
Posts from X
Most Activity
Most Activity
VIEWS2.6K
Suhail@Suhail
@bcherny Oh great!
Boris Cherny@bcherny
In the next version of Claude Code: subagents run in the background by default, so you can keep talking to Claude while your subagents work
If you want your agent to run in the foreground, just tell Claude
1hViews 2.6KLikes 2Bookmarks 0
BOOKMARKS1
Laila Rizvi@Laila_Rizvi
@bcherny Omg thank u
54mViews 10Likes 1Bookmarks 1
LIKES3
Patrick Donohoe@patrickdonohoe
@bcherny i have been spending more time in the cc desktop app recently. its fantastic for running my loops
1hViews 339Likes 3
RETWEETS1
Boris Cherny@bcherny
@HarryRead33 We fwd permission requests from subagents to your main agent, so you see what they're up to. You can also arrow down + enter to "zoom in" to a subagent and message it
36mViews 401Likes 1
REPLIES5
memorycore@memorycore
@bcherny Omg. CAn someone please help me with my account. I’ve been waiting for days.
1hViews 22Likes 2
Yheu-Shen@yheushen
@bcherny @bcherny, do we not have subagents run by default in background/parallel already? I’m already seeing this.
1hViews 391Likes 2
Boris Cherny@bcherny
@yheushen We've been rolling it out gradually over the last week to a small subset of users
1hViews 360Likes 1
Ameya@ameyaaii9dc
@bcherny How would this affect token usage and context when you talk to claude code while there are a subagents running. Can you create more tasks?
1hViews 601Likes 2
TadasG 💻@tadasgedgaudas
@bcherny okay but can you add a more comfortable code selection feature and a nicer way to review the code directly in the IDE? Similarly like Cursor has with "review next file"
thank you
1hViews 250Likes 1
Boris Cherny@bcherny
@DeepBhaleraoX Yes, just arrow down and hit enter to send messages. Separate tab coming soon (you can do it now if you opt into agent teams)
37mViews 445
JinUltimate@JinUltimate1995
@bcherny It burned my goddam tokens by running a swarm on agents on basic task.
Is this another Claude scam to waste tokens?
It's 2 days since my weekly reset on my max plan and already have 60% weekly usage.
Don't add something you can't provide enough usage for.
1hViews 105Likes 1
🥔🥔🥔@argofowl
@bcherny wen?
1hViews 126Likes 3
Boris Cherny@bcherny
@tadasgedgaudas Yes, cooking something up
1hViews 243Likes 2
Harry Read@HarryRead33
@bcherny so Claude is just going to be doing stuff while I'm not even looking now. what could go wrong
1hViews 313Likes 1
Boris Cherny@bcherny
@UnforcedAG Supported out of the box
36mViews 269Likes 1
Ahmed@ahmd3ssam
@bcherny Like++ 👌🏼
54mViews 184Likes 1
Patrick Larsen@patrick_y26
@bcherny Seriosly need more secrets on the way you work, can't get enough ltrly... makes me crazy that i can't follow up 🤣
54mViews 450
Manish@mkagenius
@bcherny Some of my agents have escaped, how do I get them back though?
1hViews 115Likes 1
Rajesh D@rajeshd102
@bcherny Exactly. We need tiered agent architectures. Is there a way to handle routine coding and reviews in the background using small, specialized edge models? They should stay completely out of the user's sight while consuming cheap local compute.
22mViews 2
Bobcat@somebobcat8327
@bcherny But can I get access to it without a Los Alamos security badge
1hViews 90Likes 1
Load more posts