From ff783a496d5dee3ed94590dac6a39a9c78903286 Mon Sep 17 00:00:00 2001
From: Aaron Sachs <898627+asachs01@users.noreply.github.com>
Date: Fri, 4 Sep 2026 12:49:27 +0000
Subject: [PATCH] docs(claude): record phase5-performance P-4 CI flake as
 sibling to SC-2

tests/integration/phase5-performance.test.ts P-4 (10 successive
write+read cycles of 100 crons, <100ms threshold) failed on a GitHub
Actions hosted runner during PR #151 CI (103.586ms; a second parallel
Unit Tests run on the same commit passed clean). The existing entry
here names phase5-performance SC-2 specifically and scopes to "the
primary dev Mac" -- this is neither: same file, same load-sensitive-
timing-assertion shape, different assertion and a cloud CI runner
instead of the local Mac. murph caught the imprecision when I first
described this as "already documented" without checking which
specific case the entry names.
---
 CLAUDE.md | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/CLAUDE.md b/CLAUDE.md
index 3e749c6576..5585c009c1 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -84,6 +84,17 @@ npm test
   fast-checker "heartbeat watchdog" fake-timer tests, phase4-performance p95
   assertions, and phase5-performance SC-2. They pass on a quiet machine —
   not regressions.
+- Same flaky class, different environment (2026-09-04, dev/murph, PR #151
+  CI): `tests/integration/phase5-performance.test.ts` P-4 ("10 successive
+  write+read cycles of 100 crons all complete in <100ms each") failed on a
+  GitHub Actions hosted runner — 103.586ms vs the 100ms threshold — while a
+  second parallel `Unit Tests` run on the same commit passed clean (run
+  33874368157, job 101027786766). Not the Mac-scoped case above (this is a
+  cloud CI runner, not the primary dev Mac) and not the named SC-2 case —
+  a sibling load-sensitive-timing assertion in the same file. Confirmed
+  unrelated to the PR's diff. Don't assume "already documented" covers a
+  new instance without checking which specific assertion and environment
+  the existing entry actually names.
 
 ## Learnings - 2026-08-04