design-review-gate

Multi-agent design reviews run by spawning five independent expert agents—PM, Architect, Designer, Security, and CTO—to validate a design artifact in parallel. This multi-agent coordination ensures comprehensive evaluation across multiple perspectives before implementation begins.

Coordinating parallel design reviews is best handled using Task Mode or Team Mode patterns. These coordination patterns balance review freshness with efficiency, allowing five independent expert agents to evaluate design quality, feasibility, and security simultaneously.

Yes, the design review gate enforces security and CTO approval by including dedicated Security and CTO reviewer agents. It applies an acceptance criterion requiring these independent expert reviews to approve the formalized brainstorming output before implementation starts.

Yes, a completed design document or brainstorming output is required. The gate activates automatically once a design artifact is ready, triggering the five parallel reviewer agents to evaluate design quality, feasibility, security, and alignment with objectives.

The multi-agent design review process iterates because enforcing an iterative review cycle with escalation ensures thorough validation. It repeats evaluations until all five independent reviewers approve the design or a human decision is reached to resolve conflicts.