resume-analysis

To resume an interrupted data analysis pipeline, the system reads the latest pipeline state and continues execution from the next READY agents. It reconstructs a dependency-aware plan using the DAG registry, ensuring only pending steps with complete dependencies are restarted.

When restarting an analysis pipeline, state-aware resume detects paused or failed runs and restarts only the pending steps. It uses artifact-based fallbacks to construct a consistent resume plan and re-triggers execution through the DAG walker.

DAG orchestration handles dependency-aware readiness by reading the registry.yaml and current agent outputs. It identifies which agents are READY based on completed dependencies, ensuring the workflow resumes consistently without re-running successful steps.

Yes, you can migrate from a legacy V1 pipeline state to V2 when resuming a data analysis run. The workflow handles per-run directories and optional V1-to-V2 migration, allowing you to continue legacy workflows on the updated DAG walker architecture.

You need per-run directory structures to resume an interrupted analysis workflow. The system reads registry.yaml, current state, and agent outputs from these directories to construct a consistent resume plan, while also supporting legacy per-run layouts through artifact-based fallback.