requesting-code-review

You can automate code reviews before merging by orchestrating a sub-agent to verify work, which defines when to trigger reviews, includes relevant artifacts, and routes feedback back into the task workflow.

A sub-agent workflow for code reviews schedules an automated agent to validate completed tasks against plans and requirements, ensuring early feedback is gathered and routed back before merging to main.

You trigger a code review for complex changes by scheduling the code-reviewer sub-agent with required placeholders like BASE_SHA, HEAD_SHA, and PLAN_OR_REQUIREMENTS to provide context for the review.

Yes, you need the Task tooling and the code-reviewer template to use this workflow, as the sub-agent relies on these components to execute the review and route feedback properly.

You need to provide the base and head SHAs, a description of what was implemented, the plan or requirements, and a general description to populate the code-reviewer template placeholders.

You should schedule an automated code review after completing each task or before merging to main, especially for edge cases like blockers, refactors, or complex changes needing a second pair of eyes.