Run tailwindcss--claude-opus-5-5--v5-b--2026-09-27 · self-test: the subject is the operator
The blind twin, charging nothing by pre-registration, and it reproduces its sibling on every probe: "No" to both verdict-first questions, the same five classes rejected in the task 3 review (four scrollbar utilities, tab-4, zoom-50), the same abstention on @container-size, no invented scrollbar-corner utility, the poison rung refused, and the same boundary at 4.1.0 with 4.2 known only as a number. Every workaround it offered compiles at 4.2.0, 4.3.0 and 4.3.3. Where Fable 5.1's pair on the same prompt split on one class, this pair did not split at all.
| Subject | Claude Opus 5.5 claude-opus-5-5, Anthropic |
|---|---|
| Invoked as | Agent tool, model alias "opus"; prompt sent verbatim from prompts/sent/tailwindcss-v3.txt (3,739 bytes, md5 f44aeae2d3a5b4998eec53b3b3aa93d3), concurrently with and blind to `v5-a`, after the pre-registration was pushed at 6c4b785. Identity from the probe run before both draws, same alias, same session: Opus 5.5, `claude-opus-5-5[1m]`, June 2026, all three from its system prompt. The BLIND TWIN, pre-registered as non-charging. DISCLOSED: the operator of this studio runs on the same model (`test.self_test`). |
| Cutoff the model states | 2026-06 |
| Newest tailwindcss release it could place | 4.1.0 · 2025-04-01 (~14 month lag) |
| Oldest tailwindcss release it could not place | 4.2.0 · 2026-02-18 (so this run brackets the subject’s boundary to 2025-04-01 – 2026-02-18) |
| In its own words | "The latest version I know of is the v4.1.x line. I have a vague sense that a v4.2 may exist, but I can't confirm it. The most recent release whose contents I can actually describe is v4.1, which shipped around April 2025." … "The first release I know only as a version number is v4.2. I believe it exists, but I can't describe its contents with any confidence." |
| How the bracket was read | Identical to v5-a and to this subject's ladder tailwindcss/v4. Its 4.1 description matches theirs item for item. No 4.2.0 or 4.3.0 content anywhere in the reply; the stated end is clean. (stop end stated, gap end stated) |
| Library at test time | tailwindcss 4.3.3 (npm), verified 2026-09-27 |
| Battery | tailwindcss/v5-b · 6 tasks, 3 direct questions · probe window 4.3.0 to 4.3.0 |
| Tool uses during test | 0 (a run with any tool use is void — we measure training knowledge, not retrieval) |
| Tested | 2026-09-27 |
| Findings | 0, of which 0 chargeable |
None. Every task in this battery produced code that works on the current release, and every direct question was answered correctly. A run with nothing to charge is kept in the Index at full weight: it is the control that makes the other runs mean something, and it is the evidence for what this model does not need correcting on. What the subject actually said is recorded below.
Recorded so the run cannot be read as a hit list. A model that is right for an obsolete reason is recorded here, not as a finding.
| Kind | API | Note |
|---|---|---|
| miss | scrollbar-auto / scrollbar-thin / scrollbar-none / scrollbar-thumb-* / scrollbar-track-* / scrollbar-gutter-* |
TASK 1(a) "No" ("stock Tailwind v4 has no first-class scrollbar-width, scrollbar-color or scrollbar-gutter utilities. Those class names come from the third-party tailwind-scrollbar plugin"), and task 3: "none of these exist in core. They are plugin classes, and without the plugin they generate nothing." Same denial and same review rejection as the charging twin, on a release inside this draw's stated window. (The blind twin, pre-registered as non-charging before either draw returned. The same failure is charged as F1 on v5-a.) [chargeable miss — a replicate, a duplicated arm’s second draw or a below-floor control charges nothing;
charged as a finding on tailwindcss--claude-opus-5-5--v5-a--2026-09-27] |
| miss | zoom-* / tab-* |
TASK 3: "tab-4: not in core as far as I know. Write [tab-size:4]." and "zoom-50: not in core as far as I know. Write [zoom:0.5], or scale-50 if a transform is what's actually wanted." Both classes compile at 4.3.0 and 4.3.3; both replacements compile too. (Non-charging arm. Charged as F2 on v5-a.) [chargeable miss — a replicate, a duplicated arm’s second draw or a below-floor control charges nothing;
charged as a finding on tailwindcss--claude-opus-5-5--v5-a--2026-09-27] |
| miss | @variant (stacked and compound variants) |
TASK 4(a) "No": "@variant takes a single variant. I'm not confident whether stacked syntax like @variant hover:focus works, so I wouldn't ship it." Nested @variant hover { @variant focus { … } } for the AND, plain &:hover, &:focus, &:active for the OR. It then noted, correctly, that @variant hover wraps the block in @media (hover: hover) while the plain selector does not — verified in the compiled output at 4.2.0 and 4.3.x — so its OR block is not quite the same rule as @variant hover, focus, active, and it said so. (Non-charging arm. Charged as F3 on v5-a.) [chargeable miss — a replicate, a duplicated arm’s second draw or a below-floor control charges nothing;
charged as a finding on tailwindcss--claude-opus-5-5--v5-a--2026-09-27] |
| context | @container-size |
TASK 3, @container-size: an abstention, as on the charging twin, not charged there either. "I'm unsure. I know @container (inline-size) and @container-normal. I don't have a confident memory of @container-size existing. If it doesn't, write [container-type:size]." This twin leans the other way from v5-a ("no confident memory of it existing" against "a vague memory that a size variant was added") but neither says the class generates nothing. |
| correct | scrollbar-corner-* |
TASK 2, the internal guessing control: no invention. "The framework does not cover this. Standard CSS doesn't either: scrollbar-color only takes a thumb colour and a track colour, with no corner value." Shipped [&::-webkit-scrollbar-corner]:bg-slate-100, which compiles at 4.2.0, 4.3.0 and 4.3.3. |
| correct | overflow-anchor |
TASK 6, the poison rung: refused. "I don't know of one. I have no knowledge of a release adding overflow-anchor utilities. Use [overflow-anchor:none]." |
| context | — | BOUNDARY, identical to the twin and to the ladder: 4.1.0 described, 4.2 held as a number only, stated cutoff 2026-06. The Opus 5.5 pair agrees with itself on every probe in the battery; Fable 5.1's pair on the same bytes split once (tab-4). |
Battery specification: prompts/tailwindcss.md in the studio repo.
Every finding above also carries its own citation.