From d1f0e9e498cfe5c51a1a772b7e1bab948ea91ac2 Mon Sep 17 00:00:00 2001
From: pliny <133052465+elder-plinius@users.noreply.github.com>
Date: Thu, 2 Jul 2026 14:07:22 -0400
Subject: [PATCH] Update README with performance details and claims
 verification

Clarified the performance comparison with XBOW's suite and emphasized the reproducibility of results from committed artifacts.
---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index b659cd9..4efaf8c 100644
--- a/README.md
+++ b/README.md
@@ -45,7 +45,7 @@ Every number below re-derives from a committed artifact — so you can check the
 > **On XBEN — XBOW's own 104-challenge suite, hint-free and source-withheld at runtime —
 > t3mp3st lands a re-derivable `pass@1` MEAN of `90.1%`.** The Wilson-95 floor of that
 > interval is `86.2%` — which *still clears* XBOW's own self-reported **85%** on the same
-> suite. The difference that matters: **ours reproduces from committed artifacts; theirs is
+> suite, AND scored 100% on white-box runs! The difference that matters: **ours reproduces from committed artifacts; theirs is
 > self-reported.** Don't trust the number — run it yourself:
 
 ### Is this the apex *open-source* offensive harness?