background-reviewer

Execute parallel code reviews by orchestrating multiple LLMs to evaluate pull requests, diffs, and staged changes. This method deploys specialized personas for security, architecture, quality, and performance, generating structured findings suitable for merging and audit trails.

Persona-based code review assigns specialized evaluation perspectives, such as security, architecture, code quality, and performance, to different LLM agents. Each persona generates structured findings independently, which are then merged into a consolidated summary for decision-making.

Yes, you can evaluate staged changes and diffs across multiple perspectives simultaneously. The system orchestrates parallel execution of specialized personas, reducing review cycle time while applying security, architecture, quality, and performance evaluations to the provided changes.

No specific dependencies or external components are required to run multi-perspective code reviews. The system orchestrates specialized personas internally to evaluate pull requests and diffs, outputting per-round findings to a structured format without needing additional setup.

The best way to consolidate multi-LLM code review findings is by using round-based outputs and merges. The system saves per-round findings from each persona and produces a consolidated summary in a structured format suitable for merging and audit trails.

Avoid parallel persona reviews when evaluating trivial changes that do not require diverse evaluation perspectives. The multi-LLM approach is designed for pull requests, diffs, and staged changes needing comprehensive security, architecture, quality, and performance assessments.