-
-
Notifications
You must be signed in to change notification settings - Fork 0
Closed
Labels
blocked-by:#236Blocked by issue #236Blocked by issue #236phase:46-checkpoint-resume-long-running-pipePhase 46: Checkpoint & Resume for Long-Running PipelinesPhase 46: Checkpoint & Resume for Long-Running PipelinesreliabilityError handling and retry logicError handling and retry logicux
Description
Description
When mgw:run starts for an issue, check if a checkpoint exists with a pipeline_stage beyond new. If found, display the checkpoint state to the user and offer: (R)esume from checkpoint, (F)resh start (discard checkpoint), (S)kip issue. Resume loads the checkpoint state and jumps to the appropriate pipeline step, skipping already-completed work.
Acceptance Criteria
- Add resume detection and auto-resume to mgw:run startup
Milestone Context
- Milestone: v8 — Agent Reliability & Failure Recovery
- Phase: Phase 46: Checkpoint & Resume for Long-Running Pipelines
- GSD Route: plan-phase
Dependencies
implement-checkpoint-writes-at-each-pipeli
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
blocked-by:#236Blocked by issue #236Blocked by issue #236phase:46-checkpoint-resume-long-running-pipePhase 46: Checkpoint & Resume for Long-Running PipelinesPhase 46: Checkpoint & Resume for Long-Running PipelinesreliabilityError handling and retry logicError handling and retry logicux