From 1421c641e8a8ab41d5e2dc1cecad537759017b26 Mon Sep 17 00:00:00 2001
From: garrytan <garrytan@gmail.com>
Date: Wed, 30 Sep 2026 20:04:19 +0000
Subject: [PATCH] feat(qa-evidence): refuse evidence observed on an older input
 snapshot than the latest capture

When native probe output declares a top-level input snapshot, materialize
compares each evidence row with the latest capture's snapshot and refuses
stale rows unless they are classified superseded, naming the captures to
rerun. ship-exploratory-late-input kept reporting a pre-change adverse probe
green after the input changed.
---
 lib/qa-evidence.ts       |  8 ++++++++
 test/qa-evidence.test.ts | 18 ++++++++++++++++++
 2 files changed, 26 insertions(+)

diff --git a/lib/qa-evidence.ts b/lib/qa-evidence.ts
index 80d8a2f9c..9e92142eb 100644
--- a/lib/qa-evidence.ts
+++ b/lib/qa-evidence.ts
@@ -264,6 +264,14 @@ function materialize(root: string, source: string) {
     const captured = readQaCapture(root, row.capture);
     return { command: row.command, contract: row.contract, expected: row.expected, classification: row.classification, observed: captured.observed };
   });
+  const snapshotOf = (observed: unknown) => object(observed) && typeof observed.snapshot === 'string' ? observed.snapshot : undefined;
+  const latestCapture = latestCompleteCapture(root);
+  const currentSnapshot = latestCapture ? snapshotOf(readQaCapture(root, latestCapture).observed) : undefined;
+  const superseded = currentSnapshot === undefined ? [] : annotations.evidence.filter((row: any, index: number) => {
+    const snapshot = snapshotOf(evidence[index].observed);
+    return snapshot !== undefined && snapshot !== currentSnapshot && row.classification !== 'superseded';
+  }).map((row: any) => row.capture);
+  if (superseded.length) throw new QaEvidenceError(`Superseded evidence: capture ${superseded.join(', ')} observed an older input snapshot than the latest capture ${latestCapture}; rerun the affected probe on current inputs, or classify the row "superseded" and keep its contract open`);
   const missing = completeCaptures(root).filter(capture => !captures.has(capture)
     && !annotations.limits.some((limit: string) => new RegExp(`\\b${capture}\\b`).test(limit)));
   if (missing.length) throw new QaEvidenceError(`Invalid report annotations: add an evidence row for capture ${missing.join(', ')} (every complete capture needs one, or name it in limits with why it is withheld)`);
diff --git a/test/qa-evidence.test.ts b/test/qa-evidence.test.ts
index b496383a1..0c8f1f1a0 100644
--- a/test/qa-evidence.test.ts
+++ b/test/qa-evidence.test.ts
@@ -315,3 +315,21 @@ test('both QA helpers answer --help with usage and exit 0, and the declared inte
   expect(qaCommandAllowed('bun bin/gstack-qa-evidence --version')).toBe(false);
   expect(qaCallerCommandAllowed('bun /abs/host/runtime/bin/gstack-qa-evidence --help')).toBe(true);
 });
+
+test('materialize refuses evidence whose declared input snapshot predates the latest capture', () => {
+  const f = fixture();
+  const command = (id: string, input: string) => `bun gstack-qa-evidence capture ${f.root} ${id} --timeout-ms 4000 -- probe ${input}`;
+  expect(f.capture('001', `console.log(JSON.stringify({ snapshot: 'before', charter: 'adverse' }))`).status).toBe(0);
+  expect(f.run('checkpoint', f.root, '001', '001', command('001', 'adverse'), 'The input changed, so the happy path must be rechecked on current inputs next.', command('002', 'happy')).status).toBe(0);
+  expect(f.capture('002', `console.log(JSON.stringify({ snapshot: 'after', charter: 'happy' }))`).status).toBe(0);
+  const rows = [
+    { capture: '001', command: command('001', 'adverse'), contract: 'README.md', expected: 'rejects', classification: 'pass' },
+    { capture: '002', command: command('002', 'happy'), contract: 'README.md', expected: 'doubles', classification: 'pass' },
+  ];
+  f.json('annotations.json', { revision: 'fixture-revision', limits: ['Adverse coverage predates the input change.'], evidence: rows });
+  const stale = f.run('materialize', f.root, 'annotations.json');
+  expect(stale.status).toBe(2);
+  expect(receipt(stale.stderr).message).toContain('capture 001 observed an older input snapshot than the latest capture 002');
+  f.json('annotations.json', { revision: 'fixture-revision', limits: ['Adverse coverage predates the input change.'], evidence: [{ ...rows[0], classification: 'superseded' }, rows[1]] });
+  expect(f.run('materialize', f.root, 'annotations.json').status).toBe(0);
+});