[BUG] TDD Orchestrator crashes with "JavaScript heap out of memory" error

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?

The TDD orchestrator agent repeatedly crashes with a heap allocation failure.

Environment:

• Platform: Windows (win32)
• Date: 2025-10-03

donet-developer.md
tdd-orchestrator.md
unit-test-writer.md

What Should Happen?

I'd like to have an orchestrator coordinate TDD between 2 agents

Error Messages/Logs

FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory

  Stack Trace:
  1: 00007FF767EDA62B node::SetCppgcReference+19819
  2: 00007FF767E4A468 v8::base::CPU::num_virtual_address_bits+93320
  3: 00007FF76688C4991 v8::Isolate::ReportExternalAllocationLimitReached+65
  4: 00007FF76688AE0E8 v8::Function::Experimental_IsNopFunction+1336
  5: 00007FF76870FAC0 v8::Platform::SystemClockTimeMillis+659328
  6: 00007FF76870CB48 v8::Platform::SystemClockTimeMillis+647176
  7: 00007FF768721E5A v8::Platform::SystemClockTimeMillis+733978
  8: 00007FF7687226D7 v8::Platform::SystemClockTimeMillis+736151
  9: 00007FF768730FCF v8::Platform::SystemClockTimeMillis+795791
  10: 00007FF7683F16A5 v8::CodeEvent::GetFunctionName+116773
  11: 00007FF70891AAFA

  GC Log:
  Multiple Mark-Compact GC cycles showing allocation failures and scavenge failures:
  - [26040:00001F8A0C14F60] 134421 ms: Mark-Compact 4015.4 (4137.2) -> 4005.0 (4142.9) MB, 2003.70 / 0.00 ms (average mu = 0.169, current mu = 0.013)
  allocation failure; scavenge might not succeed
  - [26040:00001F8A0C14F60] 138152 ms: Mark-Compact 4022.0 (4143.9) -> 4011.7 (4149.2) MB, 3701.45 / 0.00 ms (average mu = 0.069, current mu = 0.008)
  allocation failure; scavenge might not succeed

Steps to Reproduce

I was having it perform TDD for the following unit tests

• Add new test fixtures for activation/deactivation scenarios:
• When_activating_asset fixture with tests:
• Then_should_activate_asset_and_raise_event_given_asset_is_inactive
• Then_should_not_raise_event_given_asset_is_already_active
• When_deactivating_asset fixture with tests:
• Then_should_deactivate_asset_and_raise_event_given_asset_is_active
• Then_should_not_raise_event_given_asset_is_already_inactive

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.0.5

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

_No response_