peruse

Automated code review pipelines integrate detectors, reviewers, and validators to streamline quality assurance. This Skill orchestrates that process by running a Stage 0 detector, concurrent reviewers, and validators, then aggregating their findings into a final verdict.

Automated review pipelines handle security testing by invoking concurrent reviewers, including dedicated security and adversarial reviewers. These specialized reviewers run alongside standard checks to identify vulnerabilities and edge cases before the pipeline aggregates findings into a JSON output.

Automated review pipelines can be used with features having a pending review status. The pipeline executes conditionally based on this pending_review status and task completion, ensuring reviews only run when prerequisites are met before managing state transitions.

Managing state transitions after an automated code review is handled by writing review results to a specified JSON output. The pipeline then processes the aggregated verdict to determine and trigger the appropriate state transition for the reviewed feature.

Running a comprehensive plan-vehicle review in the mumei framework involves invoking a single command to start the process. This triggers the orchestration of the Stage 0 detector, concurrent reviewers, and validators to execute the full automated review pipeline.

Automated review pipelines skip validation for completed tasks due to built-in conditional logic. Execution depends on both the feature's pending_review status and whether task completion criteria are met, preventing redundant reviews on already validated features.