Cross-Team Quality Gates

Individual quality loops optimize one agent's output. Cross-team quality gates ensure that the handoff between agents does not degrade the work. QUILL produces a post at 8.7/10 quality. RENDER designs the visual assets. If RENDER's work scores 8.9/10 independently but the combined post+visual package scores 7.2/10, the integration layer has a quality problem that neither individual loop caught.

Cross-team gates evaluate the assembled artifact, not the components. The evaluation dimensions shift: instead of individual quality, the gate measures coherence (do the parts tell the same story?), consistency (do the visual and textual tone match?), and completeness (are all required elements present and properly integrated?). We run cross-team gates on every customer-facing deliverable. The integration quality score averages 8.34/10 — 0.47 points lower than the average component score of 8.81/10. That 0.47-point gap is the integration tax.