[jake-b-omni]

The OMN-20223 dashboard walk found Runs, Recent runs and Last run showing Status and Cost Not recorded on all 23 runs of a fresh onex dashboard install. Each receipt said success and $0.00. There were two causes, both specific to the local store:

• Status: the SQLite store has no boolean type, so it serves quality_gate_passed as 1. statusOf accepted only true/false.
• Cost: a run's cost came only from its delegation.savings.v1 session, and the local store serves none of those.

Change (src/pages/LocalDashboardPage.tsx)

• servedBoolean reads true/false (Postgres) and 1/0 (SQLite). Anything else, null included, is still Not recorded.
• costOf takes the session's local_cost_usd when one is served, else the decision row's own cost_usd, which feat(OMN-20753): the decisions exposure serves run status and cost omnimarket#3588 adds to the decisions exposure. A cost never measured is served null and stays Not recorded, never $0.00.
• Every per-run cost now prints in dollars with at least two decimals and is never rounded: 0 shows $0.00 and 0.0021 shows $0.0021. This matches the Spend card. Five existing assertions changed only to add the $.

Tests

• New: LocalDashboardPage.status-cost.test.tsx. Its rows have the walk's local shape: quality_gate_passed: 1, no sessions.
  • Runs shows passed and $0.00 for a measured zero cost.
  • A null cost is Not recorded, never $0.00.
  • 0 reads as failed, and null matches neither passed nor failed.
  • Last run matches Runs.
  • A served session still supplies the cost.
  • Four of the five fail against dev's LocalDashboardPage.tsx.
• Full suite: 2172 passed, 15 skipped. tsc --noEmit and eslint are clean.

Evidence-Source: OCC#13359
Evidence-Ticket: OMN-20753

🤖 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 5 changed file(s):

• src/pages/LocalDashboardPage.omn19980.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/LocalDashboardPage.runs.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/LocalDashboardPage.status-cost.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/LocalDashboardPage.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/LocalDashboardPage.tsx: 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).