chore(superagent): disable contributor-trust scoring, keep the security scan

.github/superagent.yml

@@ -3,13 +3,23 @@
 prScan:
   enabled: true
 
+# Contributor-trust scoring is DISABLED (2026-07-29). Two independent reasons:
+#
+#  1. It does not discriminate here. Established repeat contributors with dozens of merged PRs were scored
+#     "dangerous" (10/100) on volume/velocity alone, with no real signal behind it. A `trustedAuthors`
+#     allowlist was added to fix exactly that and did NOT work — shin-core was listed and still flagged
+#     `action_required` on PR #9816 — so it has been REMOVED rather than carried along inert and
+#     misleading. An allowlist that does not allow is worse than none: it looks handled while every
+#     affected PR still stalls.
+#  2. Its verdict has no consumer left. LoopOver now lists this check under `gate.ignoredCheckRuns`
+#     (#9813) on all three gate repos, so a non-passing result no longer gates, pends, or holds anything.
+#     Leaving it enabled only produces a permanently-red check on contributor PRs that nothing acts on —
+#     pure noise that makes good contributors think they have failed something.
+#
+# The SECURITY SCAN above stays enabled and still gates CI normally: that is the protection worth having,
+# and it is a different check from the same app.
 contributorTrust:
-  enabled: true
-  blockBelowScore: 30
-  # These are established, high-volume repeat contributors (dozens of merged PRs each) who kept tripping
-  # the trust scorer on volume/velocity alone, not on any real signal. Exempts them from contributor-trust
-  # scoring entirely going forward.
-  trustedAuthors: [dhgoal, shin-core, andriypolanski]
+  enabled: false
 
 comments:
   mode: detailed