diff --git a/design/prose-tests.md b/design/prose-tests.md
index 1970bcd02..ab7be189c 100644
--- a/design/prose-tests.md
+++ b/design/prose-tests.md
@@ -174,20 +174,28 @@ testing.
   A failure is a finished result; diagnosing it is a separate, human-led
   act. This is stated in prose in each definition rather than relied on
   from tool restrictions, which an agent may hold regardless.
-- **P9 — a case starts where a session starts, and runs until state
-  lands.** The only real openings are `workflow-start` (the user's), a
-  `workflow-*-entry` skill (what a bridge plan file invokes after a
-  context clear), and `workflow-discovery` (epic continuation). Navigation
-  and processing skills are always reached mid-session; a reference is
-  never entered directly. A walk carries only the context it accumulates,
-  so one begun in the middle is judged under conditions the prose was
-  never written for, and its verdict means nothing in either direction.
-  Enforced in `lib/cases.cjs` by the required `entry` field, not left to
-  authoring discipline. The same rule ends a case: stop at a gate and the
-  delta is empty, the claims turn into descriptions of a display, and the
-  asserter is left comparing the walker's account of what it showed
-  against the case's account of what it should have shown — with no
-  independent ground truth anywhere in the loop.
+- **P9 — a case starts where a session starts, and stops where its
+  claims can be answered.** The only real openings are `workflow-start`
+  (the user's), a `workflow-*-entry` skill (what a bridge plan file
+  invokes after a context clear), and `workflow-discovery` (epic
+  continuation). Navigation and processing skills are always reached
+  mid-session; a reference is never entered directly. A walk carries only
+  the context it accumulates, so one begun in the middle is judged under
+  conditions the prose was never written for, and its verdict means
+  nothing in either direction. Enforced in `lib/cases.cjs` by the required
+  `entry` field, not left to authoring discipline.
+
+  The far end is governed by the claims, not by the clock or by state. A
+  handoff is a perfectly good ending: the recorded actions are as
+  deterministic as a world delta, so which gates were read, what they
+  returned, and what was never called are all hard facts at that point.
+  What is *not* allowed is a claim with no deterministic answer — a claim
+  about text put on screen rests on the walker's account alone, since no
+  hook observes an emission. Regulate the claims and the stop point falls
+  out of them. (An earlier draft demanded every case run until state
+  landed. That was wrong twice over: it treated file changes as the only
+  hard evidence, and it would have pushed short entry walks into needing
+  conversation stubs, trading live prose for hand-written fiction.)
 
 - **P10 — what code can decide, code decides.** A case declares
   invariants over the recorded actions (`lib/invariants.cjs`) that run
diff --git a/tests/prose/cases/discussion-entry-seeds-from-carrier/case.json b/tests/prose/cases/discussion-entry-seeds-from-carrier/case.json
index 514a9adfa..ac5027dd2 100644
--- a/tests/prose/cases/discussion-entry-seeds-from-carrier/case.json
+++ b/tests/prose/cases/discussion-entry-seeds-from-carrier/case.json
@@ -6,6 +6,14 @@
     "skills/workflow-shared/references/ensure-discovery-item.md"
   ],
   "invariants": {
-    "engine_before_write": true
+    "engine_before_write": true,
+    "calls_include": [
+      "manifest get pay.discussion.pay status"
+    ],
+    "calls_exclude": [
+      "topic start",
+      "topic reopen",
+      "manifest set"
+    ]
   }
 }
diff --git a/tests/prose/cases/investigation-entry-seeds-from-carrier/case.json b/tests/prose/cases/investigation-entry-seeds-from-carrier/case.json
index 8f64940f7..64a64307a 100644
--- a/tests/prose/cases/investigation-entry-seeds-from-carrier/case.json
+++ b/tests/prose/cases/investigation-entry-seeds-from-carrier/case.json
@@ -6,6 +6,18 @@
     "skills/workflow-investigation-entry/references/invoke-skill.md"
   ],
   "invariants": {
-    "engine_before_write": true
+    "engine_before_write": true,
+    "calls_include": [
+      "manifest get crash-fix.investigation.crash-fix status",
+      "render phase-note"
+    ],
+    "calls_in_order": [
+      "manifest get crash-fix.investigation.crash-fix status",
+      "render phase-note"
+    ],
+    "calls_exclude": [
+      "topic start",
+      "topic reopen"
+    ]
   }
 }
diff --git a/tests/prose/cases/planning-entry-asks-for-late-context/case.json b/tests/prose/cases/planning-entry-asks-for-late-context/case.json
index 8ff061076..0ea0dade9 100644
--- a/tests/prose/cases/planning-entry-asks-for-late-context/case.json
+++ b/tests/prose/cases/planning-entry-asks-for-late-context/case.json
@@ -11,6 +11,20 @@
     "continue \u2014 nothing has changed since the specification"
   ],
   "invariants": {
-    "engine_before_write": true
+    "engine_before_write": true,
+    "calls_include": [
+      "render entry-gate pay.planning.pay",
+      "manifest list",
+      "knowledge.cjs query"
+    ],
+    "calls_in_order": [
+      "render entry-gate pay.planning.pay",
+      "knowledge.cjs query"
+    ],
+    "calls_exclude": [
+      "knowledge.cjs index",
+      "knowledge.cjs remove",
+      "topic reopen"
+    ]
   }
 }
diff --git a/tests/prose/cases/review-entry-fresh/case.json b/tests/prose/cases/review-entry-fresh/case.json
index a365e2cd2..6bfcb1428 100644
--- a/tests/prose/cases/review-entry-fresh/case.json
+++ b/tests/prose/cases/review-entry-fresh/case.json
@@ -6,6 +6,13 @@
     "skills/workflow-review-entry/references/validate-phase.md"
   ],
   "invariants": {
-    "engine_before_write": true
+    "engine_before_write": true,
+    "calls_include": [
+      "render entry-gate pay.review.pay"
+    ],
+    "calls_exclude": [
+      "topic reopen",
+      "topic start"
+    ]
   }
 }
diff --git a/tests/prose/cases/spec-entry-from-discussion/case.json b/tests/prose/cases/spec-entry-from-discussion/case.json
index 95045863f..484e517cc 100644
--- a/tests/prose/cases/spec-entry-from-discussion/case.json
+++ b/tests/prose/cases/spec-entry-from-discussion/case.json
@@ -8,6 +8,15 @@
     "skills/workflow-specification-entry/references/invoke-skill.md"
   ],
   "invariants": {
-    "engine_before_write": true
+    "engine_before_write": true,
+    "calls_include": [
+      "render entry-gate pay.specification.pay"
+    ],
+    "calls_exclude": [
+      "topic reopen",
+      "topic supersede",
+      "manifest set",
+      "manifest apply"
+    ]
   }
 }
diff --git a/tests/prose/cases/spec-entry-from-investigation/case.json b/tests/prose/cases/spec-entry-from-investigation/case.json
index c48f2332e..2c38f89d1 100644
--- a/tests/prose/cases/spec-entry-from-investigation/case.json
+++ b/tests/prose/cases/spec-entry-from-investigation/case.json
@@ -7,6 +7,15 @@
     "skills/workflow-specification-entry/references/invoke-skill.md"
   ],
   "invariants": {
-    "engine_before_write": true
+    "engine_before_write": true,
+    "calls_include": [
+      "render entry-gate crash-fix.specification.crash-fix"
+    ],
+    "calls_exclude": [
+      "topic reopen",
+      "topic supersede",
+      "manifest set",
+      "manifest apply"
+    ]
   }
 }
diff --git a/tests/prose/cases/start-lists-active-work/case.json b/tests/prose/cases/start-lists-active-work/case.json
index 8cd90c53f..93fc70255 100644
--- a/tests/prose/cases/start-lists-active-work/case.json
+++ b/tests/prose/cases/start-lists-active-work/case.json
@@ -5,6 +5,14 @@
     "skills/workflow-start/SKILL.md"
   ],
   "invariants": {
-    "engine_before_write": true
+    "engine_before_write": true,
+    "calls_include": [
+      "engine.cjs boot",
+      "gateway.cjs"
+    ],
+    "calls_in_order": [
+      "engine.cjs boot",
+      "gateway.cjs"
+    ]
   }
 }
diff --git a/tests/prose/lib/invariants.cjs b/tests/prose/lib/invariants.cjs
index d7243e458..d9dcbd631 100644
--- a/tests/prose/lib/invariants.cjs
+++ b/tests/prose/lib/invariants.cjs
@@ -18,9 +18,14 @@
 //   "invariants": {
 //     "engine_before_write": true,      // no .workflows write out of nowhere
 //     "calls_include": ["task init"],   // these commands must have run
-//     "calls_exclude": ["task start"]   // these must not have
+//     "calls_exclude": ["task start"],  // these must not have
+//     "calls_in_order": ["a", "b"]      // and these in this sequence
 //   }
 //
+// A check that could not have failed reports N/A rather than PASS. A
+// green tick for "there was nothing to examine" is how a corpus comes to
+// look covered while enforcing nothing.
+//
 // Deliberately not a pinned call sequence. A recorded-and-replayed
 // sequence would freeze whatever the walker happened to do, which is how
 // a test starts certifying broken prose instead of catching it.
@@ -41,7 +46,7 @@ function isWrite(row) {
   return row.tool === 'Bash' && SHELL_WRITE.test(row.detail);
 }
 
-const NAMES = ['engine_before_write', 'calls_include', 'calls_exclude'];
+const NAMES = ['engine_before_write', 'calls_include', 'calls_exclude', 'calls_in_order'];
 
 /** The commands the walk ran, in order. */
 function commands(rows) {
@@ -60,7 +65,11 @@ function engineBeforeWrite(rows) {
   );
   const firstWrite = rows.findIndex(isWrite);
   if (firstWrite === -1) {
-    return { ok: true, detail: 'no workflow state was written' };
+    // Nothing was written, so there was nothing for this check to catch.
+    // Reporting that as a pass would dress absence of coverage up as
+    // coverage — which is exactly how a corpus of read-only cases came to
+    // look green while enforcing nothing at all.
+    return { ok: true, vacuous: true, detail: 'no workflow state was written' };
   }
   if (firstEngine === -1) {
     return { ok: false, detail: `wrote ${rows[firstWrite].detail} having never called the engine` };
@@ -90,6 +99,31 @@ function callsExclude(rows, forbidden) {
     : { ok: true, detail: `ran none of: ${forbidden.join(', ')}` };
 }
 
+/**
+ * Order carries meaning a presence check cannot: a gate read after the arm
+ * it was supposed to select proves the arm was chosen some other way. The
+ * declared commands must appear as a subsequence — other calls may fall
+ * between them, but never out of sequence.
+ */
+function callsInOrder(rows, sequence) {
+  const ran = commands(rows);
+  let at = 0;
+  for (const wanted of sequence) {
+    const found = ran.findIndex((c, i) => i >= at && c.includes(wanted));
+    if (found === -1) {
+      const seen = sequence.slice(0, sequence.indexOf(wanted));
+      return {
+        ok: false,
+        detail: seen.length
+          ? `"${wanted}" never ran after ${seen.map((s) => `"${s}"`).join(' → ')}`
+          : `"${wanted}" never ran`,
+      };
+    }
+    at = found + 1;
+  }
+  return { ok: true, detail: `ran in order: ${sequence.join(' → ')}` };
+}
+
 /**
  * Run a case's declared invariants against its recorded actions.
  * Returns one result per declared check, in declaration order.
@@ -106,6 +140,9 @@ function check(rows, declared) {
   if (declared.calls_exclude && declared.calls_exclude.length) {
     results.push({ name: 'calls_exclude', ...callsExclude(rows, declared.calls_exclude) });
   }
+  if (declared.calls_in_order && declared.calls_in_order.length) {
+    results.push({ name: 'calls_in_order', ...callsInOrder(rows, declared.calls_in_order) });
+  }
   return results;
 }
 
@@ -113,7 +150,10 @@ function check(rows, declared) {
 function format(results) {
   if (!results.length) return null;
   return results
-    .map((r) => `${r.ok ? 'PASS' : 'FAIL'}  ${r.name} — ${r.detail}`)
+    .map((r) => {
+      const verdict = r.ok ? (r.vacuous ? 'N/A ' : 'PASS') : 'FAIL';
+      return `${verdict}  ${r.name} — ${r.detail}`;
+    })
     .join('\n');
 }
 
@@ -130,11 +170,15 @@ function declarationErrors(declared) {
   if ('engine_before_write' in declared && typeof declared.engine_before_write !== 'boolean') {
     errors.push('engine_before_write must be true or false');
   }
-  for (const key of ['calls_include', 'calls_exclude']) {
+  for (const key of ['calls_include', 'calls_exclude', 'calls_in_order']) {
     if (!(key in declared)) continue;
     const value = declared[key];
     if (!Array.isArray(value) || value.some((v) => typeof v !== 'string' || !v.trim())) {
       errors.push(`${key} must be an array of non-empty strings`);
+      continue;
+    }
+    if (key === 'calls_in_order' && value.length < 2) {
+      errors.push('calls_in_order needs at least two commands — one has no order');
     }
   }
   return errors;
diff --git a/tests/scripts/test-prose-invariants.cjs b/tests/scripts/test-prose-invariants.cjs
index 8179495f1..867abaaa6 100644
--- a/tests/scripts/test-prose-invariants.cjs
+++ b/tests/scripts/test-prose-invariants.cjs
@@ -141,6 +141,64 @@ describe('calls_include / calls_exclude', () => {
   });
 });
 
+describe('calls_in_order — presence is not sequence', () => {
+  it('passes when the declared calls appear in sequence', () => {
+    const rows = [
+      bash(`${ENGINE} render entry-gate pay.planning.pay`),
+      bash(`${ENGINE} manifest list`),
+      bash('cd . && node .claude/skills/workflow-knowledge/scripts/knowledge.cjs query x'),
+    ];
+    const [result] = invariants.check(rows, {
+      calls_in_order: ['render entry-gate pay.planning.pay', 'knowledge.cjs query'],
+    });
+    assert.equal(result.ok, true);
+  });
+
+  it('fails when they ran in the wrong order — the arm was chosen some other way', () => {
+    const rows = [
+      bash('cd . && node .claude/skills/workflow-knowledge/scripts/knowledge.cjs query x'),
+      bash(`${ENGINE} render entry-gate pay.planning.pay`),
+    ];
+    const [result] = invariants.check(rows, {
+      calls_in_order: ['render entry-gate pay.planning.pay', 'knowledge.cjs query'],
+    });
+    assert.equal(result.ok, false);
+    assert.match(result.detail, /"knowledge.cjs query" never ran after "render entry-gate/);
+  });
+
+  it('tolerates other calls falling between them', () => {
+    const rows = [bash(`${ENGINE} a`), bash(`${ENGINE} unrelated`), bash(`${ENGINE} b`)];
+    assert.equal(invariants.check(rows, { calls_in_order: ['a', 'b'] })[0].ok, true);
+  });
+
+  it('fails when a declared call never ran at all', () => {
+    const rows = [bash(`${ENGINE} a`)];
+    const [result] = invariants.check(rows, { calls_in_order: ['a', 'b'] });
+    assert.equal(result.ok, false);
+    assert.match(result.detail, /"b" never ran after "a"/);
+  });
+
+  it('needs two commands to mean anything', () => {
+    assert.match(invariants.declarationErrors({ calls_in_order: ['a'] })[0], /at least two/);
+  });
+});
+
+describe('a check that could not fail says so', () => {
+  it('reports N/A, not PASS, when nothing was written to examine', () => {
+    const out = invariants.format(
+      invariants.check([bash(`${ENGINE} boot`)], { engine_before_write: true }),
+    );
+    assert.match(out, /^N\/A {3}engine_before_write/);  // padded to align with PASS/FAIL
+    assert.ok(!out.includes('PASS'), 'absence of coverage must not read as coverage');
+  });
+
+  it('still reports PASS when it genuinely had something to check', () => {
+    const rows = [bash(`${ENGINE} task init pay pay`), wrote('./.workflows/a.md')];
+    const out = invariants.format(invariants.check(rows, { engine_before_write: true }));
+    assert.match(out, /^PASS {2}engine_before_write/);
+  });
+});
+
 describe('declaration', () => {
   it('runs every declared check, and only those', () => {
     const rows = [bash(`${ENGINE} boot`)];