[BUG] Task tool incorrectly reports sequential execution when run_in_background=false despite actual parallel execution

Resolved 💬 3 comments Opened Jan 29, 2026 by Leannnnnnn Closed Feb 28, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

When launching multiple subagents using the Task tool with run_in_background=false in a single message, Claude incorrectly claims that the tasks will execute
sequentially. However, the actual runtime logs show that all tasks execute in parallel.

Claude's incorrect statement:
"Note: You set run_in_background=false, which means they will execute sequentially rather than truly in parallel. If you want true parallel execution, you need to set
run_in_background=true."

What Should Happen?

Either:

  1. If parallel execution is correct behavior: Claude should NOT claim that run_in_background=false causes sequential execution when multiple Task calls are made in a

single message

  1. If sequential execution is intended: The runtime should actually enforce sequential execution when run_in_background=false

The documentation and Claude's explanation should accurately reflect the actual behavior.

Error Messages/Logs

# Claude's response before execution:
  "Now testing parallel launch of 4 subagents (note: you set run_in_background=false,
  which means they will execute sequentially rather than truly in parallel)."

  # Actual task execution log shows parallel completion:
  4 Task agents finished
  ├─ Subagent-1 · 21 tool uses · 159.0k tokens
  │  ⎿  Done
  ├─ Subagent-2 · 17 tool uses · 175.0k tokens
  │  ⎿  Done
  ├─ Subagent-3 · 27 tool uses · 159.0k tokens
  │  ⎿  Done
  └─ Subagent-4 · 32 tool uses · 128.6k tokens
     ⎿  Done

  # Total execution time: ~6 minutes
  # (Sequential execution of 4 tasks with 128k-175k tokens each would take much longer)

Steps to Reproduce

  1. Open Claude Code CLI
  2. Send a single message that invokes the Task tool multiple times (e.g., 4 Task calls) with run_in_background=false for each
  3. Observe Claude's explanation about execution behavior
  4. Compare Claude's claim with actual execution logs and timing

Example prompt:
Launch 4 subagents in parallel to analyze components 1-4.
Set run_in_background=false for all tasks.

Expected Claude behavior: Should not claim tasks will run sequentially if they actually run in parallel.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.23

Platform

Other

Operating System

Ubuntu/Debian Linux

Terminal/Shell

WSL (Windows Subsystem for Linux)

Additional Information

_No response_

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗