diff --git a/src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md b/src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md
index 5b5d4d2e6..b581d2d56 100644
--- a/src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md
+++ b/src/bmm/workflows/bmad-quick-flow/quick-spec/steps/step-04-review.md
@@ -122,13 +122,15 @@ Saved to: {finalFile}
 
 ---
 
-For best results, run **Adversarial Review** in a fresh context to ensure information asymmetry (the reviewer has no knowledge of your design decisions, only the spec):
+For best results, run **Adversarial Review** in a separate session so the reviewer only sees the spec, not your design rationale. Open a new chat/session and run:
 
 \`\`\`
 bmad-review-adversarial-general {finalFile}
 \`\`\`
 
-Once you are fully satisfied with the spec (ideally after **Adversarial Review** and maybe a few rounds of **Advanced Elicitation**), run implementation in a FRESH CONTEXT for best results:
+Bring the findings back to this session to triage and apply fixes before moving to development.
+
+Once you are fully satisfied with the spec (ideally after **Adversarial Review** and maybe a few rounds of **Advanced Elicitation**), open a new chat/session and run implementation for best results:
 
 \`\`\`
 quick-dev {finalFile}