From d914e2334466d8e71dfa45da43fe7e48c7675ec7 Mon Sep 17 00:00:00 2001
From: Teknium <127238744+teknium1@users.noreply.github.com>
Date: Wed, 4 Nov 2026 23:07:44 -0800
Subject: [PATCH] test(approval): best-of-3 both sizes and floor the bound at
 2s

Review (Greptile): a fast best-of-3 baseline with a single wide sample could put the bound under the
2s the test guide requires, so a load spike during the one wide run could fail an unchanged linear
scan. Both sizes are now best-of-3 and the bound is max(2.0, base*8).
---
 tests/tools/test_approval_pipe_shell.py | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/tests/tools/test_approval_pipe_shell.py b/tests/tools/test_approval_pipe_shell.py
index 79fa0fbadd..968489ad9c 100644
--- a/tests/tools/test_approval_pipe_shell.py
+++ b/tests/tools/test_approval_pipe_shell.py
@@ -102,9 +102,14 @@ class TestPipeRemoteToShellPattern:
             detect_dangerous_command(command)
             return time.perf_counter() - start
 
-        base = min(scan(1000) for _ in range(3))
-        wide = scan(4000)
-        assert wide < base * 8 + 0.5
+        def best(n):  # drop load spikes: the least-loaded of three samples
+            return min(scan(n) for _ in range(3))
+
+        base = best(1000)
+        wide = best(4000)
+        # Linear 4x the input stays near 4x the time; quadratic is ~16x. Floor the bound at 2s per the
+        # test guide so a fast baseline cannot make it tight on a loaded runner.
+        assert wide < max(2.0, base * 8)
 
     def test_or_list_is_not_a_pipe_for_the_widened_sink(self):
         pipe = "|"