Member exclusive access and an assertion written backwards

The verdict Kane CLI returned

The verdict points at the assertion rather than at the application:

• What Kane CLI observed - the final verification on step three did not resolve.
• The verdict - the final assertion misclassifies an available member product.
• The classification - automation_bug, config_issue, at confidence 0.93.

The step is phrased as a contrast: verify the page shows a size grid and a size guide button "rather than the members-only notice". That sentence asks for a positive state and the absence of another in one breath, and the comparison is what misfired. Step four then clicked a size and confirmed the product was buyable, which passed.

The flow under test

The first two steps are worth keeping exactly as they are. Together they prove the gate is real: a signed-out visitor sees a members-only notice with no sizes, and a valid member sign-in is accepted.

---
mode: testing
url: https://my-testing-repo-main.vercel.app/nike-clone-app?reset=true
max_steps: 45
tags: [nike, membership, gating]
---

# Nyke 8.2: Member exclusive access

Catalog objective: log in as a member and open a member-only product.
Key assertion: the product is accessible and the buy button is enabled.

## Confirm the product is gated when signed out
Click "Nyke Flyknit Elite - Member Exclusive" and verify the page shows "Members only" and a "Sign in to unlock" button instead of sizes.

## Sign in as a member
Click "Sign in to unlock", type "priya.nair@example.com" into Email, type "member2026" into Password, click "Sign in", and verify a welcome confirmation appears.

## Confirm the product unlocked
Verify the product page now shows a size grid and a "📏 Size guide" button rather than the members-only notice.

## Confirm it is buyable
Click size "US 10" and verify the "Add to Bag" button is enabled and a badge reads "Selected: US 10".

Step four is the evidence that the unlock genuinely happened. It selects US 10, confirms the Add to Bag button is enabled and checks the selection badge, and it passed. Reproduce the run with:

kane-cli testmd run tests/nike/member-exclusive_test.md

How the failure surfaced

The result file shows the failure sitting between passing steps on either side, with the step after it confirming the very state the failing step could not resolve.

---
test: ../member-exclusive_test.md
status: failed
started: 2026-09-15T10:42:41.135Z
duration_s: 122
session_id: bc92a41e-5e7a-413c-932e-5fcf2fcdcf41
---

# Nyke 8.2: Member exclusive access - Result

## Confirm the product is gated when signed out ✓ passed (34.6s)
md5: b04fc81c8762e035ad07a3906fcd4870
Click "Nyke Flyknit Elite - Member Exclusive" and verify the page shows "Members only" and a "Sign in to unlock" button instead of sizes.

## Sign in as a member ✓ passed (40s)
md5: 3ec81baf6b9a873407a1a66dd60594c5
Click "Sign in to unlock", type "priya.nair@example.com" into Email, type "member2026" into Password, click "Sign in", and verify a welcome confirmation appears.

## Confirm the product unlocked ✗ failed (43.2s)
md5: d5402035508eca37177c312610d6a3a0
Reason: Final verification failed: "the product page now shows a size grid and a "📏 Size guide" button rather than the members-only notice" - bug verdict: Final assertion misclassifies available member product [automation_bug/config_issue, confidence 0.93]
Verify the product page now shows a size grid and a "📏 Size guide" button rather than the members-only notice.

## Confirm it is buyable ✓ passed ( - )
md5: a7a8c18ac7e4280a0fd133a854a7724a
Click size "US 10" and verify the "Add to Bag" button is enabled and a badge reads "Selected: US 10".

That pattern is diagnostic. When the step after a failure proves the state the failing step was looking for, the fault is in how the question was asked, not in the answer.

Fixing the test

The fix is to split the comparison. Assert the size grid and the size guide button as a positive check, and if the absence of the members-only notice matters, make that a separate assertion rather than a clause inside the first.

Positive assertions resolve more reliably than contrastive ones, which is worth knowing before writing a suite of gating tests. How live page assertions are decided covers the mechanism, and building a suite from requirements covers phrasing checks consistently across one.