Symptom

The 2026-06-30 UX audit (finding #6) flagged a light-mode gap: the audit believed DarkBlueGrayTheme always passed ChirpyDarkColorScheme, leaving no legible onSurface/onSurfaceVariant for bright surfaces. Specifically it warned that onSurface: #AFB0B1 (the dark-scheme value, red ≈ 0.69) would be near-invisible on the light scheme’s white surface.

Root cause

ChirpyLightColorScheme was implemented in PR #401 with correct dark values for onSurface (#111111) and onSurfaceVariant (#494949), and DarkBlueGrayTheme was wired to serve it when darkTheme = false. However, KrillThemeTest had no explicit contract assertions for those two roles in the light scheme. With no guard in place, a future edit that accidentally copied dark-scheme values into the light scheme would pass all existing tests.

Fix

Added two @Test assertions to KrillThemeTest:

• light scheme onSurface is dark enough to read on a bright surface — asserts ChirpyLightColorScheme.onSurface.red < 0.25 (current value #111111, red ≈ 0.067).
• light scheme onSurfaceVariant is legible on bright surfaces — asserts ChirpyLightColorScheme.onSurfaceVariant.red < 0.50 (current value #494949, red ≈ 0.286).

No production code changed; the light scheme was already correct.

Prevention

• For every color-scheme role that the UX audit calls out by name, add a corresponding KrillThemeTest assertion so the value is under contract, not just present in the file.
• When porting values between ChirpyDarkColorScheme and ChirpyLightColorScheme, verify that any onSurface* or onBackground value has a red channel < 0.5 in the light scheme and > 0.5 in the dark scheme — opposite luminance poles.