feat: model-centric usage & spend view across providers (Scope B from #2257)#2322
feat: model-centric usage & spend view across providers (Scope B from #2257)#2322Yuxin-Qiao wants to merge 1 commit into
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0bb85adb5a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| .frame(maxWidth: .infinity, minHeight: 220) | ||
| } | ||
| } else { | ||
| let modelHostGroupID = self.controller.model.groups.first?.id |
There was a problem hiding this comment.
When local Kimi history is the only available source, KimiCodeSessionScanner emits currencyCode: "XXX" and SpendDashboardModel.currencyCode filters that out of groups, so this host-group path is never reached and the groups.isEmpty branch above shows “No local cost history yet” despite model.modelAnalysis containing Kimi rows. Kimi-only users therefore cannot see the new model history unless there is also a priced currency group; render the global model section outside currency groups or synthesize a host for token-only sources.
Useful? React with 👍 / 👎.
clawsweeper
Bot
added
proof: sufficient
|
Codex review: needs maintainer review before merge. Reviewed July 19, 2026, 6:00 AM ET / 10:00 UTC. Summary Reproducibility: not applicable. this is a new dashboard capability rather than a report of broken existing behavior. The contributor supplied direct screenshots of the changed 7-day, 30-day, and All states plus a manual macOS verification claim. Review metrics: 2 noteworthy metrics.
Merge readiness Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch. Rank-up moves:
Risk before merge
Maintainer options:
Next step before merge
Maintainer decision needed
Security Review detailsBest possible solution: Decide and document the supported aggregation contract first: either land this as clearly source-additive exact-name reporting with prominent limits, or narrow it until canonical identity and double-counting handling have a maintainer-approved design. Do we have a high-confidence way to reproduce the issue? Not applicable: this is a new dashboard capability rather than a report of broken existing behavior. The contributor supplied direct screenshots of the changed 7-day, 30-day, and All states plus a manual macOS verification claim. Is this the best way to solve the issue? Unclear. The UI and data path are coherent, but the linked discussion shows that exact-name aggregation without canonical aliases or source deduplication is a product-policy choice that needs explicit approval. AGENTS.md: found and applied where relevant. Codex review notes: model internal, reasoning high; reviewed against 1d307430a13d. Label changesLabel changes:
Label justifications:
Evidence reviewedWhat I checked:
Likely related people:
What the crustacean ranks mean
Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics. How this review workflow works
Review history (6 earlier review cycles)
|
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 18a58d70fb
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| inputs.append(SpendDashboardModel.ProviderInput( | ||
| id: "kimi:local", | ||
| provider: .kimi, | ||
| displayName: "Kimi Code CLI", | ||
| modelProviderName: ProviderDescriptorRegistry.descriptor(for: .kimi).metadata.displayName, | ||
| snapshot: snapshot)) |
There was a problem hiding this comment.
When the user clicks Refresh, the controller runs a .forceRefresh load and then a .captureOnly reconciliation; merge(outcome:capture:) only carries forced loader outputs whose ids are in forcedCodexIDs or barrierFailed, so this new kimi:local input is dropped because Kimi has no captured publication and is not a Codex account. In a Kimi-enabled dashboard, a successful manual refresh therefore removes the freshly scanned Kimi model history from loadedInputs instead of displaying it.
Useful? React with 👍 / 👎.
811bc6f to
92714de
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 92714def17
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| @State private var selectedDay: Date? | ||
|
|
||
| private var presentation: SpendModelsPresentation { | ||
| SpendModelsPresentation(analysis: self.analysis, metric: .tokens) |
There was a problem hiding this comment.
When the model data comes from a cost-only source such as Bedrock, the breakdowns carry costUSD but no totalTokens, so hard-wiring the section to .tokens makes every row value nil and presentation.points empty; the card then renders “No model-level history” even though analysis.dailyValues contains spend values. Choose .estimatedSpend when token coverage is unavailable, or expose the intended metric switch, so spend-only providers remain visible.
Useful? React with 👍 / 👎.
37f7807 to
fe82c42
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 37f78075d2
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| for line in data.split(separator: 0x0A, omittingEmptySubsequences: true) { | ||
| guard let event = try? decoder.decode(WireEvent.self, from: Data(line)), | ||
| event.type == "usage.record", | ||
| event.usageScope == "turn", |
There was a problem hiding this comment.
Kimi Code 0.14 wire.jsonl rows in the same ~/.kimi-code/sessions/.../agents/... layout can omit usageScope (see ccusage#1261), while scoped session rows are the cumulative ones to skip. With the strict event.usageScope == "turn" guard, those installs have every valid usage row ignored and scan returns nil/no local model history; please treat a missing scope as a turn record and continue excluding explicit session rows.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: fe82c42ca0
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| fileManager: FileManager = .default, | ||
| historyDays: Int = defaultHistoryDays, | ||
| now: Date = Date(), | ||
| calendar: Calendar = .current) -> CostUsageTokenSnapshot? |
There was a problem hiding this comment.
When the user’s macOS calendar is non-Gregorian (for example Buddhist), this default calendar makes CostUsageLocalDay.key write Kimi daily entries with years such as 2569. The spend dashboard later normalizes and parses provider dates with a Gregorian calendar, so those Kimi entries fall outside the current 2026 window and the local model history disappears. Please generate Kimi snapshot day keys with a Gregorian calendar that preserves the current time zone, as the dashboard model does.
Useful? React with 👍 / 👎.
fe82c42 to
ff057d5
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: ff057d5869
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| private static func normalizedDays(_ value: Int) -> Int { | ||
| value == 7 ? 7 : 30 | ||
| [7, 30, 365].contains(value) ? value : 30 |
There was a problem hiding this comment.
When settingsSpendDashboardDays is already 365 (or any caller invokes selectDays(365), which this now persists), the controller builds 365-day currency/provider totals, but the header picker in PreferencesSpendDashboardPane still only offers 7 and 30 tags. The pane can therefore open with no matching segmented selection while silently showing an unsupported 365-day overview; either keep this normalization to the picker’s 7/30 options or add the All option to that picker.
Useful? React with 👍 / 👎.
Implements Scope B from steipete#2257. - Replace per-currency "model breakdown unavailable" panel with an embedded Models card showing daily stacked usage bars and a full model ranking. - Merge models across providers by trimmed, case-insensitive exact name matching; keep token aggregation separate from cost to avoid cross-currency sums. - Add independent 7d / 30d / All range control for the model card without affecting the existing Overview 30d spend semantics. - Add Kimi Code session scanner so Kimi token-only history can contribute to the cross-provider model analysis. - Localize new UI strings across all supported languages. - Adopt upstream completeModelSummaries filtering for models list while keeping modelAnalysis built from all summaries for partial coverage. - Regenerate CodexParserHash for merged Vendored/CostUsage sources. Tests: SpendModelsPresentationTests, SpendDashboardModelTests, SpendDashboardKimiModelTests, KimiCodeSessionScannerTests, CostUsageDailyReportMergeTests, ShareStatsTests, UserFacingLocalizationCoverageTests (63 tests). Co-authored-by: Cursor <cursoragent@cursor.com>
ff057d5 to
7321949
Compare
clawsweeper
Bot
added
status: 👀 ready for maintainer look
1 participant
Summary
This PR implements Scope B from #2257: a model-centric usage & spend view embedded in the existing Usage & Spend dashboard.
Instead of the previous per-currency "Model breakdown unavailable" placeholder, users now see a Models card with:
The card sits between "By subscription" and "Daily estimated spend" and does not add a new tab, route, or Overview navigation.
Design notes
Test plan
swift test --filter "SpendModelsPresentationTests|SpendDashboardModelTests|SpendDashboardKimiModelTests|KimiCodeSessionScannerTests|CostUsageDailyReportMergeTests|ShareStatsTests|UserFacingLocalizationCoverageTests|LocalizationLanguageCatalogTests|CostUsageFetcherTests"— 106 tests pass.swift buildsucceeds.Screenshots
7d
30d
All
Fixes #2257