Scope: Fictional demo with illustrative assignments and check evidence. Unfinished work, ranked by release blockers, then recorded urgency.
Included: 17 rows, including 1 release blockers. Summary counts stay fixed when you use the display filters.
Before report filters: 1 recorded release blockers across the listed inputs. Counts cover these inputs only, not an independently certified project-wide ship gate.
Set aside as unlikely and left out: 1 (rare 1); rerun with --include-set-aside to see them.
Criteria, ranking and interpretation
{"view": "unfinished", "sort": ["blocker", "urgency"], "columns": ["id", "finding", "target", "status", "likelihood", "onestar", "ux", "owner", "next"]}
Default order: blockers first, recorded urgency, then user-impact estimate. Critical precedes High, Medium, Low and Unrated. Impact runs Severe, High, Moderate, Low, Indirect, Unrated. Effort runs Trivial, Small, Medium, Large, Unrated. Text fields sort alphabetically. Source name and line break ties. Rank numbers remain the original report rank when browser columns are sorted.
Only an unfinished THIS target blocks release. Missing targets do not become release commitments. Done-unverified remains unfinished. User-impact estimates are editorial and require a stated basis; missing ratings stay Unrated. Legacy completion words are source claims, not new verification. A status of blocked means a dependency, not automatically a release blocker.
Assignments marked Suggested are recommendations, not established responsibility. “Needs me” includes recorded and suggested user assignments. “Ready to work on” requires a documented readiness assessment; open alone does not mean ready. Last checked is an evidence-backed check date, separate from the snapshot generation time. Missing information remains Not recorded.
This is a read-only snapshot. Filters affect display only; no browser action changes the source ledgers. Zero matches does not prove release readiness.
- Fictional demonstration: assignments, next steps and check evidence come from report-annotations.json. No real work or release has been verified. P0 is the source ledger's illustrative release-blocker row.
- Source paths are relative to the repository root; source links open the adjacent sample ledger. User impact remains Unrated because no assessment is recorded.
- UNFORGET.md:16: row '----' was not read: its ID does not match the ledger ID format (verify reports it as unidentified-row).
- UNFORGET.md:118: row '----' was not read: its ID does not match the ledger ID format (verify reports it as unidentified-row).
- UNFORGET.md:149: row '----' was not read: its ID does not match the ledger ID format (verify reports it as unidentified-row).
- UNFORGET.md:175: row '----' was not read: its ID does not match the ledger ID format (verify reports it as unidentified-row).
| P0 | Example: short finding under 50 charsRelease blockerEvidence, verification & ratingsSource: UNFORGET.md · line 37 Who acts next: Coding assistant Assignment evidence: Fictional example: the project owner explicitly assigned this implementation. Next action: Implement the sample fix, then run its acceptance check. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Fictional example: scope and acceptance criteria have been reviewed; no dependency remains.
| THIS | Open | Not rated | ⚪ n/a | Unrated | Coding assistantExplicitly assigned | Implement the sample fix, then run its acceptance check. |
| P2 | Test suite: 23 flaky tests, 4 root causesEvidence, verification & ratingsSource: UNFORGET.md · line 18 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Group the failing tests by shared cause and fix one group at a time. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Fictional example: the four failure groups have reproducible test cases.
Evidence and history from the ledgerPlan with full root-cause grouping at `Documentation/Deferred/test-suite-failures.md`. Group A (race conditions) closed 2026-04-12. Re-run full suite first per Pickup Workflow. Likelihood is `measured`: the flaky tests fail in 3 of every 10 CI runs, counted over the last month. | NEXT | Open | 🔴 Probable (measured) | ⚪ n/a | Unrated | Coding assistantSuggested | Group the failing tests by shared cause and fix one group at a time. |
| P1 | Schema v3 migration paused (rollback path unclear)Evidence, verification & ratingsSource: UNFORGET.md · line 17 Who acts next: You Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Choose the rollback approach before implementation. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Agreement on the migration rollback design. Readiness evidence: Not recorded
Evidence and history from the ledgerMigration to v3 schema crashed at app launch with "Duplicate version checksums detected." V2 and V3 reference same model types -> identical checksums. Blocked: three resolution options open (fork V2, custom stage, defer entirely), none picked. Plan: `~/.claude/plans/v3-migration.md`. Files: 5 models + AppSchema + migration plan. | LATER | Blocked | Not rated | ⚪ n/a | Unrated | YouSuggested | Choose the rollback approach before implementation. |
| P4 | Search relevance overhaul (phases 1-4 done, 5-7 TBD)Evidence, verification & ratingsSource: UNFORGET.md · line 20 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Review the current search work and identify the next unfinished part. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgerBounded index row; full history here. This is the row-length split: the table cell stays a one-liner and the history lives in the block. Phases 1-4 shipped + deployed (commits `45f90604`, `38aacfca`). In progress: Phase 5 (Best Buy + PCGS sources). Pending: Phase 6 (manual fallbacks) and 26 audit findings (1 CRITICAL / 9 HIGH / 11 MEDIUM / 5 LOW from the 2026-04-01 audits). Plan: `~/.claude/plans/search-overhaul.md`. | LATER | In progress | Not rated | ⚪ n/a | Unrated | Coding assistantSuggested | Review the current search work and identify the next unfinished part. |
| P6 | Wallet pass: build server signing endpointEvidence, verification & ratingsSource: UNFORGET.md · line 22 Who acts next: You Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Confirm account access for the wallet endpoint. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Access to the sample service account. Readiness evidence: Not recorded
Evidence and history from the ledger**Spawned-from: P3.** Server endpoint, Pass Type ID, and client wiring needed before the Wallet feature can re-enable. 4-phase plan in `Documentation/Deferred/wallet-pass-archive.md` (Apple Developer Portal, Worker endpoint, Client update, Testing). Currently hidden from UI by `Sources/ViewModels/WalletViewModel.swift:244` early-return. Restore by deleting that line. **1-Star Risk 🟡 Watch:** re-enabling a payment path is exactly where a signing bug becomes a one-star review; ship it only once the endpoint is device-verified. --- | NEXT | Open | 🟡 Unlikely (reviewed) | 🟡 Watch (mid) | Unrated | YouSuggested | Confirm account access for the wallet endpoint. |
| S1 | Sentry breadcrumbs: 2 paths missing user_idEvidence, verification & ratingsSource: UNFORGET.md · line 119 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Add the missing user_id to both error breadcrumbs. Verification needed: Not recorded Last checked: 2026-04-20 Check evidence: Fictional example: a source review found both missing fields; no device check was performed. Dependencies: Not recorded Readiness evidence: Fictional example: both call sites and a focused test are identified.
Evidence and history from the ledgerSurfaced while debugging a TestFlight crash. `LoginManager.signOut()` and `OnboardingViewModel.complete()` clear the user_id from breadcrumbs before the next event fires, so subsequent crashes show empty user. Affects post-logout crash triage. | NEXT | Open | 🔴 Certain (reproduced) | ⚪ n/a | Unrated | Coding assistantSuggested | Add the missing user_id to both error breadcrumbs. |
| S3 | Empty-state copy: 3 strings hardcoded EnglishEvidence, verification & ratingsSource: UNFORGET.md · line 121 Who acts next: You Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Check all three localized strings on a device. Verification needed: Fictional acceptance check: open each affected screen in the supported locales and check wording and layout. Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledger
**`@status:done-unverified` `@verified:code`: the "done-but-owed" state.** Caught by manual scan: `Text("No items yet")`, `Text("Try a different search")`, `Text("All caught up!")` in three view files, none in `Localizable.xcstrings`. The strings are extracted and the catalog builds (code-verified), but nobody has confirmed the empty states still read right on a device. So this is NOT `done-verified` yet, and `archive` will hold it back until the device check happens. **1-Star Risk 🟢 Clear:** user-facing copy, but low exposure. The worst case is an untranslated empty state, an annoyance rather than a one-star trigger. | NEXT | Awaiting verification | Not rated | 🟢 Clear (mid) | Unrated | YouSuggested | Check all three localized strings on a device. |
| S5 | Settings: 2 toggles have no VoiceOver labelEvidence, verification & ratingsSource: UNFORGET.md · line 123 Who acts next: You Assignment evidence: Fictional example: a session logged this row on its own initiative, so only you can unpark it. Next action: Triage it: confirm a Target to unpark it, or withdraw it. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledger**`@parked`: logged by a session, waiting for you.** `NotificationToggleRow` and `SyncToggleRow` read as "switch button" with no name under VoiceOver. Parked: logged by a coding session on 2026-11-21 while fixing S1, outside that task, so nobody has triaged it yet. Revisit: the next accessibility pass, or before submitting build 14. `verify` lists it under rows waiting for you, and no session starts it until you name it or confirm its Target. **1-Star Risk 🟡 Watch (border):** an unlabeled control locks VoiceOver users out of a setting, the kind of thing an accessibility review calls out. --- | SOMEDAY | Open | Not rated | 🟡 Watch (border) | Unrated | YouSuggested | Triage it: confirm a Target to unpark it, or withdraw it. |
| A2 | radar-suite: ObservableObject + @Published (legacy)Evidence, verification & ratingsSource: UNFORGET.md · line 151 Who acts next: Alex Assignment evidence: Fictional example: Alex was assigned this review by the project owner. Next action: Review the proposed observation migration. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgerradar-suite finding ID `MOD-001`. Migration path: ObservableObject + @Published -> @Observable macro. Spec: `Documentation/Architecture/Modernization.md`. Incremental approach: convert when touching a file, don't chase the chain. Likelihood is `n/a`: a modernization, not a bug that can happen. | NEXT | Open | n/a | ⚪ n/a | Unrated | AlexExplicitly assigned | Review the proposed observation migration. |
| A4 | radar-suite: try? swallowing errors in 14 spotsEvidence, verification & ratingsSource: UNFORGET.md · line 153 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Replace the swallowed errors with explicit handling and focused tests. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgerradar-suite finding ID `ERR-003`. Convert `try?` calls that silently swallow real errors. Use `ModelContext+Logging.swift`'s `fetchWithLogging()` helper. Targets are listed in the radar report. | NEXT | Open | Not rated | ⚪ n/a | Unrated | Coding assistantSuggested | Replace the swallowed errors with explicit handling and focused tests. |
| U1 | iPad: keyboard doesn't dismiss on first sheet openEvidence, verification & ratingsSource: UNFORGET.md · line 176 Who acts next: You Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Check keyboard dismissal on the first sheet opening on iPad. Verification needed: Fictional acceptance check: open the sheet for the first time and confirm the onscreen keyboard dismisses when expected. Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgerReproduced manually 2026-04-19 on iPad Pro 13" simulator (`@status:in-progress`: a repro exists but no fix yet). First time presenting `AddItemSheet`, the keyboard appears but Done button doesn't fire; subsequent opens work. Likely SwiftUI focus-state initialization race. **1-Star Risk 🟡 Watch (border):** a first-run input trap on iPad is user-facing friction a reviewer can hit on their very first session. It sits near the At-risk border because it's on the *first* open, not a deep edge case. | NEXT | In progress | 🟡 Likely (reproduced) | 🟡 Watch (border) | Unrated | YouSuggested | Check keyboard dismissal on the first sheet opening on iPad. |
| S2 | DateFormatter cached in 4 spots (perf, not correct.)Evidence, verification & ratingsSource: UNFORGET.md · line 120 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Consolidate the four DateFormatter instances. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgerFound while reading code in PerformanceProfiler. Four call sites construct DateFormatter inside hot loops. Each construction is ~5ms. Move to lazy properties on the owning view models. | LATER | Open | Not rated | ⚪ n/a | Unrated | Coding assistantSuggested | Consolidate the four DateFormatter instances. |
| S4 | Color asset catalog has 12 unused entriesNeeds reconciliationEvidence, verification & ratingsSource: UNFORGET.md · line 122 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Compare the asset references with the ledger count before deleting assets. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded Needs reconciliation: Fictional example: a later scan reports fewer than 12 unused assets; the recorded count needs reconciliation.
Evidence and history from the ledger12 colors in `Assets.xcassets` have zero references. Cleanup is mechanical but easy to skip in routine PRs. Worth a 30-min cleanup pass before audit-readiness review. | SOMEDAY | Open | Not rated | ⚪ n/a | Unrated | Coding assistantSuggested | Compare the asset references with the ledger count before deleting assets. |
| A3 | code-review: 6 oversized files (>800 lines)Evidence, verification & ratingsSource: UNFORGET.md · line 152 Who acts next: Unassigned Assignment evidence: Not recorded Next action: Agree which large file to split first. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgercode-review-tool flagged 6 files >800 lines. None are causing problems today; tracked for future split. List: `MainView.swift` (1240), `OrderManager.swift` (910), `SearchService.swift` (864), `SettingsView.swift` (823), `ChatViewModel.swift` (812), `ReportGenerator.swift` (806). | LATER | Open | Not rated | ⚪ n/a | Unrated | Unassigned | Agree which large file to split first. |
| A5 | radar-suite: 47 async calls without .task modifierEvidence, verification & ratingsSource: UNFORGET.md · line 154 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Review cancellation behavior before planning the async-call migration. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgerradar-suite finding ID `ASY-008`. View-level async calls that should use `.task` modifier for auto-cancellation. Bulk migration; defer until a clean sprint. | SOMEDAY | Open | Not rated | ⚪ n/a | Unrated | Coding assistantSuggested | Review cancellation behavior before planning the async-call migration. |
| U2 | macOS: window position not restored after hiber.Evidence, verification & ratingsSource: UNFORGET.md · line 177 Who acts next: Coding assistant Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Trace window restoration and propose a focused fix. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Not recorded Readiness evidence: Not recorded
Evidence and history from the ledgermacOS window saves position to UserDefaults, but hibernation seems to skip the save. Cosmetic only; no data loss. Affects `WindowGroup`-based apps on macOS 14+. Rated `rare` on a `reasoned` basis, the weakest one, so no policy may set it aside without asking you. | LATER | Open | 🟢 Rare (reasoned) | ⚪ n/a | Unrated | Coding assistantSuggested | Trace window restoration and propose a focused fix. |
| U3 | TestFlight: 1 user reported chart colors look wrongEvidence, verification & ratingsSource: UNFORGET.md · line 178 Who acts next: You Assignment evidence: Fictional example: this role is proposed for the next step, not a confirmed assignment. Next action: Choose accessible chart colors with the product owner. Verification needed: Not recorded Last checked: Not recorded Check evidence: Not recorded Dependencies: Product owner feedback on the proposed palette. Readiness evidence: Not recorded
Evidence and history from the ledger**`@status:blocked`: awaiting user follow-up.** TestFlight feedback from one user, no screenshot attached. Possibly a colorblind-accessibility issue with chart series colors (currently 3 of the 5 are warm colors, which can collapse for deuteranopia). Can't proceed without the user's follow-up. **1-Star Risk 🟡 Watch:** an accessibility complaint that reaches TestFlight can reach a review; mid-band because it's one unconfirmed report, not a reproduced defect. --- | SOMEDAY | Blocked | Not rated | 🟡 Watch (mid) | Unrated | YouSuggested | Choose accessible chart colors with the product owner. |