From 8ecb75e241bb9520c239be8cb4e43ef95cbd09df Mon Sep 17 00:00:00 2001
From: Rudy Celekli <47457359+rudycelekli@users.noreply.github.com>
Date: Fri, 2 Oct 2026 14:21:49 -0400
Subject: [PATCH] Keep shadow-test failures from retrying successful production
 work

Signed-off-by: Rudy Celekli <47457359+rudycelekli@users.noreply.github.com>
---
 .../engineering-autonomous-optimization-architect.md     | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/engineering/engineering-autonomous-optimization-architect.md b/engineering/engineering-autonomous-optimization-architect.md
index 28a5fc64..07e448b4 100644
--- a/engineering/engineering-autonomous-optimization-architect.md
+++ b/engineering/engineering-autonomous-optimization-architect.md
@@ -58,7 +58,14 @@ export async function optimizeAndRoute(
       
       // Background Self-Learning: Asynchronously test the output 
       // against a cheaper model to see if we can optimize later.
-      shadowTestAgainstAlternative(serviceTask, result, getCheapestProvider(providers));
+      // Shadow failures must not enter the production retry/circuit-breaker path.
+      // Deferring the call also captures synchronous adapter failures.
+      void Promise.resolve()
+        .then(() => shadowTestAgainstAlternative(
+          serviceTask, result, getCheapestProvider(providers)))
+        .catch((error) => {
+          console.error('Shadow evaluation failed', error);
+        });
       
       return result;