Retry Behavior

If a task crashes or times out:

• The dispatcher re-queues it as a new run
• The new worker sees prior attempts in kanban_show() response
• Don’t repeat the failed path without understanding why it failed
• Common retry reasons:
  • timed_out: Reduce work scope or increase timeout
  • crashed: OOM or segfault - reduce memory footprint
  • spawn_failed: Profile config issue - ask human via kanban_block

Dispatcher Behavior