reflexion:critique

Multi-agent critique spawns parallel independent judge agents to assess requirements, architecture, and code quality, then synthesizes their findings into a coherent consensus report with actionable recommendations.

You orchestrate multi-judge consensus by spawning three parallel judge reviews that evaluate different aspects independently, followed by a debate session to resolve disagreements and produce a final consensus report.

Yes, multi-agent review workflows are designed for complex projects with diverse stakeholders, enabling parallel independent judgments across different evaluation criteria to ensure comprehensive coverage.

The best way to structure a quality-assurance critique is using a multi-agent workflow that provides explicit traceability and a reproducible consensus process through independent evaluations and synthesized debate resolutions.

Disagreements during multi-judge debate sessions are handled through structured debate sessions that facilitate reasoned resolution, ensuring the final consensus report reflects thoroughly evaluated and reconciled judgments.

You should avoid multi-agent consensus for simple tasks that do not require diverse stakeholder input or multi-dimensional evaluation, as the overhead of spawning parallel independent judges outweighs the benefits for straightforward reviews.