Claude Code v2.1.311 ·
Crashed workers are restarted straight away, with a shared wait between retries
After a worker crashes, Claude Code now tries to restart it at once, and the crash notice no longer says it is respawning
You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
1Signal: worth watching, 1 to 5
Background AgentsArea: what it touches
ImprovementsKind: in v2.1.311,
ImprovementsSection of the release
What
A worker is a separate process that runs part of Claude Code's work, such as an agent. When one crashes, Claude Code now:
- tries to start a replacement at once
- sets a shared waiting period before later retries, where it used to set a simple timer
The crash notice still reports that the worker crashed, but no longer ends with "respawning…".
Why
Crashed workers may come back sooner than before.
See this entry in the whole of v2.1.311 →
