v0.2.0
Resilience Overhaul
Major reliability improvements for long-running teams and crash recovery.
- Timeout watchdog: teammates stuck busy beyond the TTL are automatically timed out and aborted (configurable via
OPENCODE_ENSEMBLE_TIMEOUT) - Spawn rollback: if the initial prompt fails, the member, session, and DB entry are all cleaned up
- Crash recovery on plugin init: stale busy members marked as errored, orphaned sessions aborted, undelivered messages redelivered
- Shutdown resilience: abort failures handled gracefully, busy-after-shutdown triggers re-abort
Full Changelog: v0.1.2...v0.2.0