[Retracted] Windows commit exhaustion was caused by a runaway test in my own code, not Claude Code

Retracted — please disregard this report.

I originally attributed a system-wide commit-memory exhaustion on Windows to Claude Code. That was wrong. Per-process sampling later showed a single python.exe from my own test suite (a test with unbounded recursion) growing to about 285 GB in under 7 minutes. The Claude Code processes stayed at roughly 0.4-0.8 GB the whole time.

The /heapdump "High memory growth rate" figure I cited is not evidence of a leak either: in that report, bytesPerSecond is exactly RSS divided by process uptime, so it looks high for any young process.

I've deleted my earlier comment. Sorry for the noise — closing as not planned.