From 3138422b95dc716fbd73fe78c67d2d51ea76e572 Mon Sep 17 00:00:00 2001
From: John Donalson <mirlok@dr.com>
Date: Sun, 18 Oct 2026 22:51:11 -0400
Subject: [PATCH 1/2] docs: add CoIR and historical benchmark methodology
 references

---
 e2e/demo.test.ts        |  4 ++++
 src/routes/+page.svelte | 10 +++++++++-
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/e2e/demo.test.ts b/e2e/demo.test.ts
index 4fa120a9..a24d0650 100644
--- a/e2e/demo.test.ts
+++ b/e2e/demo.test.ts
@@ -17,6 +17,10 @@ test('home page explains model-agnostic intelligence and paid Starter pricing',
 		'80 paired comparisons across 40 code-navigation tasks'
 	);
 	await expect(page.locator('.evaluation-results')).toContainText('23% higher');
+	await expect(page.getByRole('link', { name: 'CoIR code-retrieval benchmark ↗' })).toHaveAttribute(
+		'href',
+		'https://github.com/CoIR-team/coir'
+	);
 	await expect(
 		page.getByRole('link', { name: 'TokensToGold methodology and scoring code ↗' })
 	).toHaveAttribute('href', 'https://github.com/noodlbox/tokens-to-gold');
diff --git a/src/routes/+page.svelte b/src/routes/+page.svelte
index a262527c..cdf1f871 100644
--- a/src/routes/+page.svelte
+++ b/src/routes/+page.svelte
@@ -363,7 +363,15 @@
 			21.8.
 		</p>
 		<p>
-			Public retrieval-benchmark reference:
+			Public retrieval-benchmark references:
+			<a href="https://github.com/CoIR-team/coir" target="_blank" rel="noopener noreferrer"
+				>CoIR code-retrieval benchmark ↗</a
+			>,
+			<a
+				href="https://github.com/Context-Engine-AI/Context-Engine/blob/3f511257382c8c7820b6748338a820a46673900f/docs/BENCHMARKS.md"
+				target="_blank"
+				rel="noopener noreferrer">our benchmark methodology (January 2026) ↗</a
+			>, and
 			<a href="https://github.com/noodlbox/tokens-to-gold" target="_blank" rel="noopener noreferrer"
 				>TokensToGold methodology and scoring code ↗</a
 			>. The Context Engine figures above come from our separate agent A/B evaluation.

From 699cecc14733e646a0f423f586a4901070d63d90 Mon Sep 17 00:00:00 2001
From: John Donalson <mirlok@dr.com>
Date: Sun, 18 Oct 2026 22:51:50 -0400
Subject: [PATCH 2/2] copy: distinguish adaptive learning from fixed benchmark
 snapshots

---
 src/routes/+page.svelte | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/src/routes/+page.svelte b/src/routes/+page.svelte
index cdf1f871..3df6ceaa 100644
--- a/src/routes/+page.svelte
+++ b/src/routes/+page.svelte
@@ -325,7 +325,7 @@
 				</div>
 				<div class="visual-feature-text">
 					<span class="visual-feature-title">Self-Learning</span>
-					<span class="visual-feature-desc">Gets smarter as you search</span>
+					<span class="visual-feature-desc">Adapts from feedback when enabled</span>
 				</div>
 			</div>
 			<div class="visual-feature">
@@ -362,6 +362,10 @@
 			from 52.2% to 64.3%. Average tokens decreased from 1.01M to 0.88M and tool calls from 25.5 to
 			21.8.
 		</p>
+		<p>
+			With learning enabled, retrieval adapts as feedback accumulates. These measurements capture
+			the tested model, configuration and sessions.
+		</p>
 		<p>
 			Public retrieval-benchmark references:
 			<a href="https://github.com/CoIR-team/coir" target="_blank" rel="noopener noreferrer"