Stage 1 rubric: tighten renders-cleanly to actual render failures

[dadachi]

Summary

Third iteration of the same fundamental issue. PR #50 dropped domain-match from the Stage 1 rubric, but the remaining renders-cleanly criterion still leaked domain-semantic judgment via its phrasing "placeholder text where real content should appear" — the vision judge interpreted "this is a welcome screen, not a queue dashboard" as failing that clause. Same domain-match question in different clothes.

Verified live

Against the post-substrate-fix iOS welcome screenshot ("Welcome to Vet Clinic Queue" with three sparkles), median-of-3 sampling oscillates between PASS and FAIL across runs. Sample FAIL rationale:

"The home screen is sparse with only generic sparkle icons and a welcome title, lacking real queue content; the layout feels unfinished/placeholder-like rather than a functional clinic queue interface."

That's Stage 2 territory per docs/SPEC.md, not Stage 1's "did it render at launch."

New criterion wording

Does the screen render without an actual rendering failure — that is, no
crash dialog, no broken-image-icon glyphs, no text overlapping other text,
no content cut off the side of the screen? A welcome / launch / onboarding
screen with decorative graphics counts as PASS as long as nothing is
technically broken; do not judge whether the screen looks "finished" or
shows the app's domain content.

Explicitly tells the judge: "decorative graphics with welcome text is PASS." Removes the "where real content should appear" clause that invited Stage-2 interpretation.

Verified post-fix

Both iOS and Android welcome screenshots from the substrate now PASS both criteria consistently:

=== ios: PASS
  no-substrate-leak: PASS — only "Welcome to Vet Clinic Queue", ...
  renders-cleanly: PASS — no overlapping text, no broken image icons, no content cut off.
=== android: PASS
  no-substrate-leak: PASS — only "Welcome to Vet Clinic Queue", ...
  renders-cleanly: PASS — no overlapping text, no broken images, no content cut off.

Sampling variance should be much lower with unambiguous criterion wording.

Test plan

• npm run ci — 19/19 green.
• Direct runLayer3 calls against the captured screenshots (iOS + Android) → both PASS both criteria.
• After merge: NATIVEAPPTEMPLATE_VISUAL=1 npm run dev -- "..." should report Layer 3 2/2 pass.

Out of scope

• Stage 2 mobile-mcp navigation past welcome to the actual domain UI. That's where domain-semantic judging belongs.

🤖 Generated with Claude Code