raise_negative_gap raised a negative gap from the interior widths only, so a first layer narrower than the gap's size let the second start before it (widths 10, 100 x5 at X 0, 80, 200, 260, 400, 460: the modal -40 gave X 0, -30, 30, ...). The second must start after the first, g > -w(0), in whole pixels floor(-w(0)) + 1. Figma kept the order in every measured run; a narrow first layer was not measured, so this bound keeps the order rather than follow a reading. Test: tidy_tests::a_narrow_first_layer_still_keeps_the_second_after_it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Model-Effort: Opus 5.5 drove and wrote
Measured in Figma's web app on 2026-11-13. Co-Authored-By below.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Model-Effort: Opus 5.5 drove; Opus 5.5 briefed; Sonnet 5.5 measured and wrote at medium
inspect.code {language: "tailwind", mobile} dropped every mobile max-width and put the desktop one behind lg:, so a fixed-size layer that is the same size on both frames (a 48 x 48 badge) stretched to the full width below 1024 px. A mobile max-width that equals the desktop one is now kept as a plain class, with no lg: copy. A differing max-width still goes fluid on mobile with the desktop cap behind lg:.
Tests: the issue's two frames (a 1440 and a 390 page, each with a 48 x 48 badge) keep max-w-[48px] unprefixed, and a page-wide banner still gets its lg: cap only. The slashed-names test gave the CTAs frame the same 342 px on both frames, which the new rule rightly keeps, so its desktop frame is now 600 px wide. docs/mcp.md notes the rule.
Model-Effort: Opus 5.5 drove; Fable 5.2 queued; Opus 5.5 briefed; Sonnet 5.5 wrote at medium
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Third Figma run (web app, 2026-11-12): the mean gap rounds half up (X 0, 60, 135, 215 gives 0, 72, 144, 216; 0, 60, 135 gives 0, 68, 136); a tie between repeated gaps takes the smaller; a row is one row group of layers and may overlap along X, a negative gap counting in the mean (X 0, 60, 105, 170 gives 0, 57, 114, 171); a grid is packed along the axis with more groups (2 x 3 by centred columns, 3 x 2 by centred rows); a diagonal, an irregular layout and layers that all overlap are unavailable. The earlier cell rule is removed.
Choices, not measured: a 2 x 2 grid as columns, auto layout flow children unavailable, the 2,000 layer cap, rotated layers by their box, spacing with two layers, no rounding with Snap to pixel grid off. Grid positions along a stack can differ by a pixel from Figma's (it read 52 and 51 for the same mean 51.67).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Model-Effort: Opus 5.5 drove; Fable 5.2 briefed; Sonnet 5.5 measured and wrote at medium
Measured in Figma's web app on 2026-11-12 (top-level rectangles): a row, a column and a 2 x 3 grid of unequal sizes, eight runs on the gap, and the cases where Tidy up is greyed out (fewer than three layers, overlapping layers). Figma keeps the order, the sizes and the first layer, evens one gap per axis (the most common existing gap, else the mean taken down to a whole number) and tidies a grid by columns.
Choices, not measured: the smaller of two equally common gaps, a mean ending in .5 going down, gaps a pixel apart differing, a diagonal being a row, a grid with an empty place, a grid always read by columns, one overlapping pair making it unavailable, auto layout flow children making it unavailable, the 2,000 layer cap, rotated layers by their box, spacing with two layers, and no rounding with Snap to pixel grid off. A selection that is none of a row, a column or a grid keeps the earlier rule unchanged.
The Design panel's Tidy up row is greyed out with a tip when unavailable; the command then moves nothing and adds no undo step.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Model-Effort: Opus 5.5 drove; Fable 5.2 briefed; Sonnet 5.5 measured and wrote at medium
Trevor BrownandClaude Sonnet 5.5
BFlat
NetterSilverandClaude Opus 5.5