feat(usage): break down usage by project

[saphid]

What Changed

The Usage page can filter totals and breakdowns to one project without combining same-named projects from different environments. Transcript working directories are resolved to stable project identities, while usage outside a project and usage whose project cannot be determined remain distinct states.

Why

Provider and model totals do not show which workspace incurred the spend. Project attribution makes that split usable without turning a repository read failure into a plausible but incorrect “outside projects” total.

Stack and Review Scope

This PR is cumulative and lands after #9014. Review the project-attribution increment; later PRs include both predecessors.

Verification

The first current-head CI Test job passed all 4,053 web tests but failed on an unhandled cancelAnimationFrame is not defined callback in upstream fileEditorHighlight.test.ts. That file is unchanged from main; its 7 focused tests passed locally. A failed-job retry was requested but GitHub rejected it because repository admin rights are required; no retry occurred. This observed teardown flake is retained as a verification limitation.

Current candidate afbb52522aa4a21c8c563854289f3a9d9d12b51f includes upstream main 39802c06117fae0b3da43624b0d54309c5437c72. The focused UsageService and chart-gesture suites passed (30 tests), and the server package typecheck passed. Repository failure doubles follow the current imported-transcript interface. Earlier evidence below retains its stated revision and scope; integrated web evidence and its attribution limits are described below.

A direct Claude Opus 5 high review attempt for this integration batch exited 1 because OAuth expired; no Claude review occurred.

Candidate: e44f88f174f0872250f0f105fe570cf3fd20d34e, including upstream main 82c2b7ffb4d450572f5baf9a70693f1e25cfdfed.

• The full focused contribution check at ba873b8 passed 128 tests across 7 files, plus the affected package typechecks.
• The final 82c2b7f delta removes an unused pricing helper: 7 pricing tests, server typecheck, Knip and git diff --check passed on the pre-repair main composition. The chart/runtime contribution was unchanged by that final delta.
• The chart merge preserves upstream's live column builder and this PR's daily/hourly and range behavior. Fresh independent source review of the shared resolution found no actionable issues.

Focused commands

vp test run apps/server/src/usage/UsageService.test.ts apps/server/src/usage/usageAggregation.test.ts apps/server/src/usage/usageScanCache.test.ts apps/server/src/usage/usageTranscripts.test.ts apps/web/src/components/usage/UsagePage.test.tsx packages/shared/src/usageMerge.test.ts apps/web/src/components/usage/UsageProviderChart.test.ts
vp run --filter t3 typecheck
vp run --filter @t3tools/web typecheck
vp run --filter @t3tools/contracts typecheck
vp run --filter @t3tools/shared typecheck
vp test run apps/server/src/usage/usagePricing.test.ts
vp run --filter t3 typecheck

Refresh correctness checks

The refresh repair passed the focused UsageService suite (10 tests), 3 chart-gesture regression tests, and 7 web usage-state tests. Server and web package typechecks, changed-file formatting/lint and diff checks passed. Independent source review found no actionable issue in this frozen repair.

These repairs cancel a chart gesture if its date window changes, let each environment progress after its own pricing request, and coalesce concurrent tokened summary scans. Thread-enabled branches await their environment summary before refreshing thread rows. The source-scan revision guard prevents older aggregation from pruning files discovered by a newer scan. Integrated web evidence and remaining client limits are described below.

UI Changes

The project breakdown separates Atlas and Beacon totals. The integration retains this PR’s project grouping; later stack changes add cache-write columns and a shared table layout.

Before: upstream main 39802c06117fae0b3da43624b0d54309c5437c72. After: cumulative #9308 integration 7e248026469ec7be6d5c2bdaccf86f380e0c2476, captured in an isolated web preview with two synthetic projects, three conversations and generated transcripts. This reuses the retained feature in the integrated stack; it is not a separate capture of this PR’s individual head. Focused tests above were run on the individual published head. The 30-day preset ends at the last complete day on the integration; main includes the current day. Both captures contain the same fourteen days of synthetic usage.

Clean breakdown recording · Annotated breakdown recording. This shows project → thread → expand → collapse; idle time is trimmed from 109.7 s to 12.57 s, without a performance claim.

Clean project-filter recording · Annotated project-filter recording. Idle time is trimmed from111.93s to7.17s; unsuccessful setup attempts before the successful selection are omitted from the viewing cut, with the raw source retained.

Limits: A supplementary filter recording selects Atlas at $45.56 and clears to all projects at $62.00. Equal-name projects across environments and outside/unknown choices are not exercised. No React Native or SwiftUI mobile capture is represented here. Clean/annotated packet validation passed; uploaded derivatives were retrieved and hash-verified.

Checklist

• The intended increment is focused; its cumulative dependency is identified above.
• I explained what changed and why.
• Current-head UI screenshots are attached; this gap is described above.
• Current filter and reverse/reset interaction proof is still required.

Implemented and reviewed with GPT-6 and GPT-5.6 Sol in the Codex harness. A direct Claude Opus 5 high availability attempt exited 1 because OAuth had expired; no Opus review occurred.