[jake-b-omni]

The OMN-20223 walk on 2026-10-08 found no Baseline spend figure on the Overview. The mockup shows it beside Actual spend and Verified savings, and the figure is already served: the same metering-summary.v1 all-time row that the Spend and Savings cards read carries counterfactual_usd, the number onex metering --json reports (0.015292 in the walk).

Change. A Baseline spend metric card binds counterfactual_usd from that row through the existing MeteringTotalCard path. It reads the same all-time row as Spend and Savings, so the three figures cannot disagree.

• required_fields match the Savings card's: baseline_model, pricing_manifest_version and baseline_state. So the card is captioned with the row's baseline model and pricing manifest, plus the "Modelled" note.
• An unresolved baseline renders Baseline unresolved, never $0, even when the row serves 0.
• Layout: the free slot at row 8, col 8, beside Run locally and Avg saving / call. The headline row is full, and reflowing it would move every panel the walk's screenshots compare.

Release. 1.1.5 → 1.1.6 (project-level lockfile fields only) with a CHANGELOG entry, so onex dashboard can pin a bundle that carries the card.

Tests. src/pages/LocalDashboardPage.omn20752.test.tsx (7) renders the Overview with the walk's all-time row. All 7 fail without the card.

• B1: the card shows $0.0153, and a different served amount shows that amount.
• B2: an unresolved baseline shows "Baseline unresolved" and never $0, whether counterfactual_usd is null or served as 0.
• B3: the caption names the baseline model, the pricing manifest and "Modelled".
• B4: no row shows a typed "Not measured", with no digits.
• B5: a day row's counterfactual is not the headline.

pages.test.ts gains counterfactual_usd in the Overview's metric keys. The lab catalogue already lists that column for metering-summary.v1. Full vitest: 2174 passed, 15 skipped. tsc and eslint are clean.

Evidence-Source: OCC#13354
Evidence-Ticket: OMN-20752

🤖 Generated with Claude Code

[onexbot-occ-writer]

OCC autobind did not mint a companion for this PR: no changed-file candidate could be proven RED against the merge base, and emitting a PR-existence probe instead would be non-falsifiable evidence (OMN-15247). Hand-authored evidence is required.

Considered 7 changed file(s):

• CHANGELOG.md: release-line source, but the diff also changes a path that is not a release artefact, so release lines are not offered
• package-lock.json: no candidate grammar reads this file type (only Python declarations, uv.lock lines, release-artefact lines and contract-pin lines are proposed)
• package.json: no candidate grammar reads this file type (only Python declarations, uv.lock lines, release-artefact lines and contract-pin lines are proposed)
• src/pages/LocalDashboardPage.omn20752.test.tsx: no candidate grammar reads this file type (only Python declarations, uv.lock lines, release-artefact lines and contract-pin lines are proposed)
• src/pages/local/overview.contracts.yaml: no candidate grammar reads this file type (only Python declarations, uv.lock lines, release-artefact lines and contract-pin lines are proposed)
• src/pages/local/overview.page.yaml: no candidate grammar reads this file type (only Python declarations, uv.lock lines, release-artefact lines and contract-pin lines are proposed)
• src/pages/local/pages.test.ts: no candidate grammar reads this file type (only Python declarations, uv.lock lines, release-artefact lines and contract-pin lines are proposed)

This verdict is occ-autobind's, derived from this head and its merge base alone. A second producer, the occ-companion-effect backstop, consumes its own command on the lab bus and may still mint a companion under the generic binding. Whether it does depends on that consumer running, not on this diff (OMN-18876).