Register pilot C "Engineered Precision": /next/pilots/precision/fractional-cto/ - #563
Merged
Conversation
…ional-cto/
Pilot C of the three-register design comparison (10.01), and the first one with
no Claude Design artifact behind it - pilots A and B each translated one, this
register is built from the research column itself ("precise, engineered,
nothing hidden"). Cool white with #fafafa panels, structural 1px borders in
#e4e4e7 / #d4d4d8, zero radius, ink #18181b. The BORDERS are the design: the
hero is two grid cells butted against one vertical rule rather than a card
floating beside the copy, the stats and case cards are cells sharing single
hairlines, and the comparison table is ruled in both directions like a spec
sheet. Inter 400/500/600/700 for prose, JetBrains Mono 400 for every label,
numeral and button.
COPY IS UNCHANGED, again - `diff` against pilot B's stub is six lines (title,
description, `register`, `font`) and nothing else. Both screenshot tests assert
the same strings as pilots A and B, so a copy edit reaching only one pilot
fails.
Ruby is the standard #cc342d and RESTRAINED to four jobs: a 6px square ahead of
each eyebrow, the comparison's active-column rule, the step numeral, and the
filled primary CTA. It never washes or tints a surface. Devices with no slot in
the anatomy were translated: the brand mark is an outlined hairline box (pilot
B's is a solid block), list markers are 1px dashes, the timeline is a numbered
protocol whose numeral gutter is divided by a full-height rule, and the leading
zero is CSS because "01" belongs to this register, not to the shared anatomy.
The accent does NOT die on the inverted band here, which is the interesting
register-fidelity result. Pilot B's deep red is 2.31:1 on ink and had to invert
to paper; #cc342d is 3.46:1 on #18181b - under the 4.5:1 text floor, over the
3:1 a filled button's edge needs, with white-on-ruby at 5.13:1 inside it - so
the closing CTA stays ruby. Measured, not assumed. Structural hairlines step up
to #71717a (3.67:1) on that band, because a register whose design IS its
geometry cannot let the geometry dissolve when the ground goes dark.
Two defects the browser caught that the source could not:
- The comparison's "theirs" column rendered in FULL INK. `.rr-table tbody td`
scores 0,1,2 on its two tag names and outranks the template's `.rr-td-muted`
at 0,1,0, so the comparison stopped comparing while the CSS still read
correct. Caught by walking computed colours per row, not by reading the file.
The same shape is live in next-editorial.css - pilot B's muted column is
actually its body tone. Fixed here by scoping the override past the default.
- A `gap:1px over a line-coloured ground` grid shows that ground as an empty
cell whenever the last row is partial: `auto-fit` collapses only tracks empty
in EVERY row, so three stats in two phone-width tracks left a grey block.
Single column below 640px.
- (Also: the month-one panel shipped with padding-right: 0 and its bullets
flush to the panel edge. Found by looking at the screenshot.)
Gates, with the instruments proven rather than trusted. bin/hugo-build clean.
bin/rake test:unit 292 runs, 6164 assertions, 0 failures. The pilot-glob gates
in next_rail_test.rb already reach this page and it was PROVEN by smoothing
this pilot's quote and watching the failure name
next/pilots/precision/fractional-cto/index.html. PurgeCSS set-diff empty across
56 selectors on a PRODUCTION build, and an injected orphan `.rr-th-probe` rule
was stripped from the bundle (byte-identical hash), so the diff bites; the
positional column selectors and the muted-column fix both survive the purge.
Computed font-weight walk: 123 elements, zero synthesized - Inter 400 (39) /
500 (5) / 600 (14) / 700 (1) and JetBrains Mono 400 (63), every pair backed by
a loaded face. Contrast walk 122 elements at 1440 AND 390, zero failures, worst
passing 4.63:1, and the walk was proven live by an injected 1.67:1 probe that
it caught alone. All four buttons clear their non-text edge (3.46-5.13:1) and
label (5.13-17.72:1). Zero console messages, zero non-2xx, zero third-party
hosts - JetBrains Mono is REUSED from pilot B's binary rather than duplicated.
No horizontal overflow at 390; the table scrolls in its own box (348 visible /
680 content); h1 lands at 36px mobile, 60px desktop. Scoped desktop + mobile
screenshots green against these committed macOS baselines, with pilots A and B
still green in the same run (8 runs, 6 screenshots, 0 failures).
Known and NOT introduced here: the full critical suite has 9 pre-existing stale
macOS baseline failures (the #540 recolour backlog item), and qtest escalates to
--all for this change because the new static font files are unmapped.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011SP5gaqXEgUie8pdFrmbeJ
Contributor
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
Screenshot diffs detected
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Pilot C — the third and final register of the design comparison (10.01). Same skeleton, same copy as merged pilots A and B; cool-technical skin: Inter (self-hosted variable latin) + JetBrains Mono (reuses pilot B's binary — no duplicate), pure white ground ruled by 1px structural borders, spec-sheet table with mono parameter labels, ruby
#cc342das restrained accent (active-column marker, step numerals, primary CTA only).The controlled comparison is now complete and verified at DOM level: all three pilots render 553 identical text tokens (one md5), 398 identical body tags, identical class sequences — one anatomy, three skins, nothing else varies.
Preview:
/next/pilots/precision/fractional-cto/.Review chain
opus author → adversarial opus verifier → SHIP with zero blocking findings. The author found and fixed three render-level defects pre-review (muted-column specificity, auto-fit grid ghost cell at 390, panel padding) — each verified live by the reviewer with revert-probes (e.g. the grid defect measured at 22,913px² exposed ground pre-fix vs 2,101px² shipped). The reviewer also caught two of its own instrument errors (sed anchor swallowing the hero; computed font-weight can't prove non-synthesis — replaced with advance-width measurement: 1285.7/1302.4/1319.0/1335.6px, four distinct weights from one variable binary).
Cross-pilot finding (follow-up PR queued, not this one)
The author's specificity defect exists on master in BOTH pilots A and B:
.rr-table tbody td(0,1,2) outranks.rr-td-muted(0,1,0), so their comparison columns render undifferentiated. Verifier confirmed with computed colors and a generalized detector (A=1, B=1, C=0 hits). One-line scope fix per file + baseline re-records, next PR.Gates
23-string banned sweep zero hits (rendered, all three pilots) · blockquote verbatim-slice of
data/testimonials.yaml(glob gate proven to bite on this page) · tenure derived · contrast walk 122 elements × 2 viewports, 0 failures, probe-verified · unit 292 runs, 6164 assertions, 0 failures · scoped screenshots 8/8 green · PurgeCSS set-diff empty, probe-verified · zero console errors, zero third-party hosts · noindex + sitemap-absent.Notes
next/keys.🤖 Generated with Claude Code
https://claude.ai/code/session_011SP5gaqXEgUie8pdFrmbeJ