dcr-lead-init

To lead a deep code review round, initialize the session by binding a specific task_name to scope the review, then coordinate reviewer findings and manage carry-forward artifacts through human-in-the-loop decision points across multiple rounds.

Managing code review artifacts across multiple rounds requires maintaining a live round packet and a findings_ledger to carry forward classifications and verification steps, ensuring explicit task association and focused human-guided coordination throughout the lifecycle.

A live findings ledger works by capturing and tracking reviewer findings, classifications, and verification steps across active review rounds, allowing the human-guided lead to coordinate carry-forward artifacts and produce human-facing execution plans.

Yes, explicit task binding is required to start a code review lead session. You must invoke the runtime with a distinct task_name at startup to bind the lead to the specific task scope and ensure round lifecycle adherence.

Yes, you can coordinate multiple reviewers within a single code review round by classifying their findings, managing the live round packet, and routing them through human-in-the-loop decision points to verify results and generate execution plans.

A human-guided lead for deep code review rounds should not be used when the process requires full automation without human-in-the-loop decision points, as this approach relies on explicit participant coordination and live round context management to function correctly.