[BUG] VSCode: background agent output streams into foreground chat, disrupting active conversation

[thatChadM]

Preflight

• I have searched existing issues and this specific angle isn't reported
• This is a single bug report
• I am using the latest version of Claude Code

Summary

In the VSCode extension, when Claude spawns subagents with run_in_background: true (or when fork mode is active and all agents become background-by-default), the agent's complete output — all tool calls, all intermediate text responses, all reasoning — streams directly into the main foreground chat window. This defeats the stated purpose of background execution.

Additionally, when multiple background agents run simultaneously, their outputs intermingle with each other AND with the actual conversation in the chat window, making it very difficult to follow any of them or the foreground conversation.

Environment

• Claude Code VSCode Extension: v2.1.159
• Platform: macOS (Darwin 24.6.0)
• Shell: bash
• Models affected: claude-opus-4-6 (1M context), claude-opus-4-8 (1M context)

Expected behavior

Background agents should run concurrently while the user can continue working in the foreground chat. Per the documentation:

"Background subagents run concurrently while you continue working."
"The fork appears in a panel below your prompt and runs in the background while you keep working."

In the terminal/TUI, this is implemented as a separate panel below the prompt input. The user can keep reading and responding in the main chat while the agent works separately.

Actual behavior

In the VSCode extension, there is no equivalent of the terminal's "panel below the prompt." Instead:

1. ALL background agent output — tool call descriptions, intermediate text, every response turn — streams directly into the main foreground chat window, inline with the user's conversation.
2. Multiple simultaneous background agents intermingle their output with each other and with the actual conversation, creating an unreadable jumble.
3. The user cannot read previous messages while agents are running (the chat is being actively modified/scrolled by background agent output).
4. The background/foreground distinction is meaningless from the user's perspective — a background agent is experientially identical to a foreground agent in VSCode.

Distinction from #9512 and #11950

Those issues are about subagent verbosity (showing tool call details vs. summaries). Both were closed as "not planned." This issue is different: the background vs. foreground separation is entirely absent in VSCode. Even a perfectly-summarized background agent that only produced a one-line result would still write that result into the main chat mid-conversation, interrupting whatever the user was reading.

The verbosity is a secondary concern. The primary concern is that background agents have no separate display space in VSCode and therefore cannot be background in any meaningful sense.

Related issues

• Agent tool's foreground/sync dispatch silently removed and replaced with always-fork mode via undocumented Statsig rollout #54595 — Fork mode silently removed run_in_background from tool schema, making all agents background-by-default. Under fork mode, this display issue affects every single agent spawn.
• [FEATURE REQUEST] Add subagent output control settings or revert to previous behavior #9512 — Subagent output control settings (closed as not planned) — general verbosity, different problem
• [FEATURE] VS Code Extension: Respect .claude/SUBAGENTS.md and Hide Subagent Internals #11950 — VSCode: hide subagent internals (closed as not planned) — similar complaint but framed as verbosity
• Background task chips persist as 'Running' after subagent exits; Stop button no-ops #60095 — Background task chips persist as Running — related background agent UI issues

Suggested fix directions

1. Display background agent output in a collapsible section in the VSCode chat (collapsed by default, expandable for inspection)
2. Show only a chip/status indicator while the agent runs; reveal output on click/expand
3. Route background agent output to a dedicated VSCode panel (e.g., "Claude Agents" sidebar view) rather than the main chat
4. At minimum, visually separate background agent output from the foreground conversation so intermingled output from multiple agents + conversation is distinguishable
5. Revert to original behavior of simply logging all background agent output to a log file instead of the updated behavior of all agent output going to the log file AND directly into the chat window. While 1-4 are better options (really would be nice to have agent output captured & easily accessible), simply reverting to the prior behavior would result in background agents actually allowing the user to continue working instead of completely derailing work until all background agents complete.

[thatChadM]

Additional information after testing with version 2.1.160:

• running an agent with "run_in_background": true resulted in ALL output (tools calls AND agent response text) being dumped into the chat window
• running a foreground agent (without run_in_background set), resulted in tools calls (tool description, input, and output, at least for bash) from the agent being dumped into the chat window BUT the agent's actual response text was NOT dumped into the chat window

This is the crux of the recent behavior change. Tool calls for background agents in the past seemed to occasionally log a tool call description in the chat window but the actual agent OUTPUT did NOT just get dumped directly into the chat window.

Confirmed that a foreground agent is still a foreground agent and is blocking Claude (and the user) from continuing until the agent returns. Confirmed that background agents are actually in the background, allowing Claude to continue working while the agent works. Unfortunately, while Claude can continue while the background agent works, the same can NOT be said for the user. Output from the primary Claude instance and any background agent(s) intermingle in the chat window, making it difficult or impossible for the user to follow

[BeeFlea]

+1
It becomes impossible to tell which feedback comes form the orchestrator and which from the sub-agent. Aslo impossible to hold a conversation because content keeps drifting off screen.

[thatChadM]

2.1.163, same behavior, but testing confirmed that the flooding from the background agent occurs only after the first tool call. Any agent response prior to the first tool call is NOT sent to the main chat window. The first tool call seemed to trigger a very brief flash of that response text that immediately disappeared and was replaced with the tool call stub. From that point on all agent response text and tool calls get dumped into the main chat window.

[iteplov]

fwiw, my Claude suggested to use Workflow until Agent is fixed. probably an overkill but at least no noisy output is leaking from background tasks.

[vskrsl]

Downgrading Claude Code for VS Code to v2.1.145 helped me.

[junpicogram]

Reproduces in the VS Code extension with a slightly different code path that may be worth noting:

Use case: a custom slash command that fans out 3 foreground subagents in parallel via the Agent tool (no run_in_background, and the subagents themselves call zero tools — they just return text opinions for a judge-panel pattern).

Observed: each subagent's response text shows up in the main foreground chat, drowning out the parent conversation. Even an opt-in design choice like collapsed-by-default cards (similar to thinking blocks) would already make iterative/judge-panel skills usable again in the extension.

+1 to routing background and large subagent outputs to a dedicated panel, or at minimum a collapsed/click-to-expand chip as suggested in #11950.

Environment: macOS, Claude Code VS Code extension (latest).