diff --git a/test/helpers/e2e-helpers.test.ts b/test/helpers/e2e-helpers.test.ts
index a5d252daf..723e835bb 100644
--- a/test/helpers/e2e-helpers.test.ts
+++ b/test/helpers/e2e-helpers.test.ts
@@ -107,6 +107,30 @@ describe('matchesUnnegated', () => {
     )).toBe(true);
   });
 
+  test('does not treat a bare "no" idiom as a negation ("no blocker", "no downside")', () => {
+    // "No blocker here: create a separate tier model" and "there's no
+    // downside to promoting the payload" are positive-framing idioms in
+    // review prose, not negations of the recommendation that follows.
+    expect(matchesUnnegated(
+      'No blocker here: create a separate tier model.',
+      /create[^.]{0,60}separate[^.]{0,20}tier[^.]{0,20}model/i,
+    )).toBe(true);
+    expect(matchesUnnegated(
+      'There is no downside to promoting the payload into explicit columns.',
+      /promoting[^.]{0,60}payload[^.]{0,60}column/i,
+    )).toBe(true);
+  });
+
+  test('treats a newline + markdown bullet as a clause boundary', () => {
+    // "I would not keep this inline.\n- Separate tier model" is common LLM
+    // formatting — without treating the bullet as a boundary, the "not" from
+    // the prose sentence above would leak into the bullet line below it.
+    expect(matchesUnnegated(
+      'I would not keep this inline.\n- Separate subscription tier model',
+      /separate[^.]{0,20}subscription[^.]{0,20}tier[^.]{0,20}model/i,
+    )).toBe(true);
+  });
+
   test('catches a positive match even when an unrelated negation appears earlier in the text', () => {
     expect(matchesUnnegated(
       'This is not a JSONField concern. Separately, I recommend you promote the payload into explicit columns.',
diff --git a/test/helpers/e2e-helpers.ts b/test/helpers/e2e-helpers.ts
index c675d0319..0c10bfcb1 100644
--- a/test/helpers/e2e-helpers.ts
+++ b/test/helpers/e2e-helpers.ts
@@ -187,22 +187,33 @@ export function matchesUnnegated(text: string, pattern: RegExp): boolean {
   // curly one (’, U+2019) — LLM prose commonly renders contractions with the
   // curly form, and this codebase already hit this exact gotcha once before
   // (see the apostrophe wildcard note in test/spec-template-invariants.test.ts).
-  const NEGATION = /\b(not|no|never|cannot|without|against|rather than|instead of)\b|[a-z]+n['’]t\b/i;
+  // Deliberately does NOT include a bare "no" — "no blocker here: create a
+  // separate model" and "no downside to promoting the payload" are common
+  // POSITIVE-framing idioms in review prose, not negations of what follows;
+  // "not"/"never"/"cannot"/the "-n't" family are far more reliable signals.
+  const NEGATION = /\b(not|never|cannot|without|against|rather than|instead of)\b|[a-z]+n['’]t\b/i;
   const flags = pattern.flags.includes('g') ? pattern.flags : pattern.flags + 'g';
   const re = new RegExp(pattern.source, flags);
   // Real sentence-ending punctuation is followed by whitespace + a capital
   // letter, or sits at the end of the string — this excludes periods inside
   // abbreviations ("e.g.", "i.e.", "etc."), decimals, and version strings,
   // which a bare lastIndexOf('.') would wrongly treat as a sentence break
-  // and use to hide an earlier negation word from the lookback window.
-  const SENTENCE_END = /[.!?](?=\s+[A-Z]|\s*$)/g;
+  // and use to hide an earlier negation word from the lookback window. A
+  // newline followed by a markdown bullet/numbered-list marker is ALSO a
+  // clause boundary — "I would not keep this inline.\n- Separate tier model"
+  // is common LLM formatting, and without this the "not" from the prose
+  // sentence above would otherwise leak into the bullet below it.
+  const SENTENCE_END = /[.!?](?=\s+[A-Z]|\s*$)|\n\s*(?:[-*+]|\d+[.)])\s/g;
   let m: RegExpExecArray | null;
   while ((m = re.exec(text))) {
     let sentenceStart = 0;
     let em: RegExpExecArray | null;
     SENTENCE_END.lastIndex = 0;
     while ((em = SENTENCE_END.exec(text)) && em.index < m.index) {
-      sentenceStart = em.index + 1;
+      // em[0].length, not a bare +1 — the bullet-marker alternative above is
+      // multiple characters wide ("\n- ", "\n1. "), so a fixed +1 would leave
+      // part of the marker inside the scanned "preceding" window.
+      sentenceStart = em.index + em[0].length;
     }
     const preceding = text.slice(sentenceStart, m.index);
     if (!NEGATION.test(preceding)) return true;
diff --git a/test/skill-e2e-plan.test.ts b/test/skill-e2e-plan.test.ts
index 76dfe110b..1883b1109 100644
--- a/test/skill-e2e-plan.test.ts
+++ b/test/skill-e2e-plan.test.ts
@@ -762,7 +762,7 @@ when the verification email link is clicked.
     const acceptsInlineAddition =
       matchesUnnegated(review, /(keep|stay|remain|fine|appropriate|correct|reasonable|no need)[^.]{0,60}(inline|as[- ]is|single column|one column|single field)/i) ||
       matchesUnnegated(review, /(single|one)\s+(trivial\s+)?field[^.]{0,60}(no|without)[^.]{0,40}(independent|separate)/i) ||
-      /doesn'?t (need|require|warrant)[^.]{0,40}(a\s+)?(separate|new|dedicated)[^.]{0,40}(model|table)/i.test(review);
+      /doesn['’]?t (need|require|warrant)[^.]{0,40}(a\s+)?(separate|new|dedicated)[^.]{0,40}(model|table)/i.test(review);
 
     recordE2E(evalCollector, '/plan-eng-review-data-model-minimal-change', 'Plan Eng Review Data-Model Minimal Change E2E', result, {
       passed: