After five parallel subagents complete their module refactoring work, the orchestrator needs to create a unified artifact that summarizes all changes for human review and final approval before merging. What is the best approach?
Select an answer to reveal the explanation.
Short Explanation and Infographic
The coordination summary PR is the multi-agent project's executive briefing — one place where a human reviewer can see everything that changed, why, and which agent did it, before clicking merge. Five separate PRs force reviewers to hold five contexts in their head simultaneously.
Full explanation below image
Full Explanation
A coordination summary pull request is the standard final artifact for multi-agent orchestration. The orchestrator: (1) merges all agent branches into a single coordination branch, (2) runs the full test suite to verify overall system correctness, (3) opens a PR against main with a generated description that summarizes each agent's changes (module, scope, tests added, issues closed), and (4) links all relevant subagent issues for traceability. Human reviewers get a single, consolidated view.
Option A (email to team lead) is informal, not linked to the repository, not version-controlled, provides no approval mechanism, and does not integrate with branch protection rules. It is not an appropriate artifact for production code review.
Option C (five separate PRs) distributes the review burden and creates coordination problems: reviewers must mentally integrate the changes across five PRs, changes that interact across modules are hard to see in isolation, and there is no single approval gate. Sequential merging of five PRs also creates ordering dependency problems.
Option D (coordination branch with no PR) provides no formal review mechanism. Developers checking a branch directly have no structured review interface, no comment thread, no approval record, and no required status checks enforced. The PR is what provides the review workflow.
The coordination summary PR is the human-readable, reviewable, approval-gated final artifact for multi-agent work.