mav-claude-code-recovery

To recover Claude Code workflows after a crash, you need state persistence and crash recovery mechanisms. This involves defining state formats and persisting progress after each milestone to enable reliable session resumption.

Subagent failure isolation prevents cascading issues during autonomous coding by containing errors within individual subagents. This approach supports autonomous recovery and stops a single subagent failure from disrupting the entire workflow.

The best way to ensure artefact durability across workflow restarts is immediate posting of designs, plans, and status updates. This practice protects against data loss and keeps humans informed during interruptions.

Claude Code subagent orchestration loses progress when interrupted due to reliability gaps in state persistence. You must enforce disciplined failure handling and persist progress after each phase to maintain workflow resilience.

Yes, you need clearly defined state formats for crash recovery in autonomous coding sessions. Defined formats allow the system to enforce state persistence and execute robust retry logic during unexpected interruptions.

Yes, you can use state persistence to resume interrupted Claude Code subagent tasks. By persisting progress after each milestone, the workflow can reliably resume from the last saved state without losing completed work.