requesting-code-review

Automate code reviews by dispatching a reviewer subagent after each task or major feature. You provide a description of what was implemented, the plan reference, and the base and head SHAs to trigger a structured diff evaluation before merging to main.

A structured code review requires a correctly filled description of what was implemented, a plan reference, and the base and head SHAs. This precise context enables the reviewer subagent to evaluate the diff thoroughly without exposure to session history.

Trigger a code review in a subagent-driven workflow after completing each individual task, after finishing major features, or before merging to main. It is also recommended to dispatch a review when refactoring or debugging to catch regressions early.

Yes, this code review automation works with Git branches by requiring the base and head SHAs to identify the exact changes. Supplying these SHAs allows the reviewer subagent to evaluate the specific diff between the two states in your repository.

Context preservation during automated code review works by routing only the relevant work product context to the reviewer subagent. It avoids exposing the entire session history, ensuring the reviewer focuses solely on the provided diff and implementation details.

The best way to catch regressions before handoffs is to request a structured code review before merging a feature branch. This surfaces potential gaps and ensures alignment with requirements by routing precise context to a reviewer subagent.