refine

Backlog preparation for overnight execution is automated by orchestrating Clarify, Research, and Spec phases to lock in intent and scope. This produces research.md and spec.md artifacts, then sets the backlog status to refined after user approval.

To lock in requirements before an overnight run, the refine process runs Clarify to establish intent, Research to explore solutions, and Spec to finalize requirements. It then registers the produced artifacts in the cortex lifecycle index for downstream traceability.

Yes, user approval is required before a backlog item status is set to refined. The workflow produces spec.md and research.md artifacts during the Clarify, Research, and Spec phases, but the status only updates to refined after you explicitly approve the output.

Research and spec artifacts are registered in cortex/lifecycle/index.md. The refine process generates cortex/lifecycle/{lifecycle-slug}/research.md and spec.md, logging them in the index to enable downstream automation and traceability for the overnight workflow.

The best way to coordinate Clarify, Research, and Spec for a lifecycle slug is using the refine orchestration flow. It sequentially runs these phases to produce research.md and spec.md under cortex/lifecycle/{lifecycle-slug}/, ensuring scope is locked before overnight execution.

Overnight workflows need refined backlog items to ensure intent and scope are locked before execution begins. The refine process produces spec.md and research.md artifacts, registering them in the cortex lifecycle index to provide downstream automation with traceable, approved requirements.