[paynejd]

What changed and why

Implementer feedback on TBv3: the results-table header row was white like the rows under it, so the columns were hard to pick out.

• Header row background is now the design token surface.s90 (#e3e0f9), applied as bgcolor: 'surface.s90' on the header row of TableResults (concepts, mappings, references, repos, orgs, users) and of the grouped references table (ReferenceSourceGroupedResults), which shared the same header style. Labels were already bold (<b>, weight 700) and stay that way.
• A 1px surface.nv80 rule above and below the header (the outline token TBv3 already uses for dividers). The bottom rule is on the header cells. The top rule is the results toolbar's existing bottom border, which changes from rgba(224, 224, 224, 1) to the same token. This is a judgment call beyond the ticket's colour decision. The selected row's highlight (primary.90, #e2dfff) and surface.s90 are almost the same value, and the toolbar's selection tint is close too. Without the rules, a selected first row and a toolbar with a selection merge into one lavender block with the header (compare the zoomed "selected" shots below). If you'd rather separate them by colour, the selected-row highlight would need its own decision.
• Other tables (dialogs, attribute and version tables, associations) are unchanged.

The design-system page that documents this treatment is in OpenConceptLab/ocl-design-system (linked in a comment below once it's open).

How I verified it

• Ran the dev build locally against the public API with a read-only proxy, and drove it with headless Chromium. In every state the header cells compute to rgb(227, 224, 249) with labels at weight 700: global search, CIEL concepts, a selected row, a scrolled table, grouped references, and the ungrouped references table with the translation toggle.
• Checked that the select-all checkbox (empty, indeterminate), the sort arrow on hover, the Raw/Translation switch, and the sticky header while scrolling are all legible on the tint.
• eslint --ext .jsx,.js src/ is clean, and npm test passes (20 of 20). No unit test was added, because the change is styling only.

Screenshots

Public content only (global search for "malaria", CIEL source, CIEL/MPX collection). The screenshots/ocl_issues-2948 branch holds only these images and can be deleted after review.

	Before	After
Global search, concepts
CIEL concepts
CIEL, a shown row and two checked rows
Same, zoomed: toolbar selection, header, shown row
Sticky header while scrolled, sort arrow on hover
Collection references, grouped by source
Collection references, ungrouped, with the translation toggle

closes OpenConceptLab/ocl_issues#2948

🤖 Generated with Claude Code

https://claude.ai/code/session_01TjM9dGow2KV1Gwr2dWQdph

[paynejd]

Codex review, pass 1 (codex-cli 0.163.0, commit eccd8c7)

Verdict: clean. No findings.

findings.json

{
    "reviewer": "codex-cli 0.163.0",
    "perspective": "defects",
    "mode": "full",
    "commit": "eccd8c7f",
    "verdict": "clean",
    "prior": [],
    "findings": []
}

[paynejd]

Design-system page for this treatment: OpenConceptLab/ocl-design-system#17 (TableResults page, merge after this PR).

[paynejd]

Simplification review (Claude subagent, commit eccd8c7)

Verdict: clean. Nothing to cut: each of the six style changes serves an acceptance criterion. No commits after the Codex pass, so no follow-up check was needed.

simplify.json

{
    "reviewer": "claude subagent",
    "perspective": "simplify",
    "mode": "full",
    "commit": "eccd8c7f",
    "verdict": "clean",
    "prior": [],
    "findings": []
}