Objective
Run a full native audit of text/foreground colour across AgencyZero. Analytics is the known case: some headings, values, labels, or table text appear to use an incorrect, missing, transparent, or insufficiently contrasted foreground. Intentional muted text must be distinguished from broken colour resolution.
Required detection
- Text that paints no foreground pixels despite having content and geometry.
- Transparent, invalid, unresolved, or unexpectedly inherited foreground colour.
- Foreground that effectively matches the authored surface/background.
- Contrast below the intended semantic tier in both dark and light modes.
- Drift between title, body, muted, status, cost/input/output accent, and table/header tokens.
- Text clipped or occluded in a way that presents as missing colour.
- Dynamic theme axes and custom properties producing invalid or unintended text colours.
Evidence and coverage
Use Blitz native inspection/paint evidence against real app surfaces, with Analytics as a required reproduction. Add ps-qa/audit checks that can sweep all text-bearing semantic nodes and report selector, role/name, resolved foreground, nearest painted surface, contrast, and token/source where available. Add focused regression outcomes for every confirmed defect. Do not use jsdom screenshots as proof.
Acceptance
- Dark and light mode sweeps complete with zero unexplained missing/invalid foregrounds.
- Every low-contrast finding is fixed or explicitly documented as an intentional exception.
- Analytics labels, values, descriptions, and table headings/body remain readable and use their intended semantic colour tiers.
Objective
Run a full native audit of text/foreground colour across AgencyZero. Analytics is the known case: some headings, values, labels, or table text appear to use an incorrect, missing, transparent, or insufficiently contrasted foreground. Intentional muted text must be distinguished from broken colour resolution.
Required detection
Evidence and coverage
Use Blitz native inspection/paint evidence against real app surfaces, with Analytics as a required reproduction. Add ps-qa/audit checks that can sweep all text-bearing semantic nodes and report selector, role/name, resolved foreground, nearest painted surface, contrast, and token/source where available. Add focused regression outcomes for every confirmed defect. Do not use jsdom screenshots as proof.
Acceptance