Skip to content

TEST_PLAN.md

Latest commit

 

History

History
1473 lines (1185 loc) · 77.3 KB

TEST_PLAN.md

File metadata and controls

1473 lines (1185 loc) · 77.3 KB

PathKeep Behavior Safety Net Test Plan

Date: 2026-05-26 MST

Scope: repository-wide behavior safety-net audit for the active WORK-V03-PAPER-REDESIGN-A block. The goal is mutation-resistant behavior coverage, not numeric coverage theater.

Rules For This Pass

  • One module at a time. After each module: run the targeted tests, run that module's coverage command, record coverage movement and suspected bugs, then make a checkpoint commit before continuing.
  • Product bugs are not silently fixed in this pass. If a correct behavioral expectation fails, keep the failing test evidence and report the bug for confirmation before changing production code.
  • Every added test must assert behavior that would fail under plausible mutations such as boundary flips, inverted guards, dropped error handling, stale cache reuse, or changed return values.
  • Coverage is a floor. Branches with no meaningful assertion remain gaps even if line coverage is high.

Reproducible Baseline Commands

bun run coverage:js
bun run coverage:rust

Actual baseline captured on 2026-05-26:

bun run coverage:js
Test Files  276 passed (276)
Tests       2023 passed (2023)
All files   statements 99 | branches 98 | functions 99.51 | lines 99.5

bun run coverage:rust
Rust coverage verified at 100% for 35246 instrumented source lines and
1634 source functions.

Important gate drift:

  • docs/plan/program/quality-matrix.md says bun run coverage:js requires 100% statements / branches / functions / lines.
  • vitest.config.ts currently enforces only statements 99 / branches 98 / functions 99 / lines 99.
  • The JS baseline therefore passes the current config while failing the documented shipping contract. Treat this as a safety-net gap, not as done.

Rust note: coverage:rust uses the project verifier's semantic source-line and function rules. It does not provide Rust branch coverage; branch confidence must come from focused tests plus later cargo mutants.

Module Coverage And Risk Map

Risk levels:

  • P0: data loss, privacy/security, archive mutation, large-archive performance, desktop command contract.
  • P1: user-visible workflow state, route navigation, migration/import/export, AI/provider-gated behavior.
  • P2: presentational components or fully pinned deterministic helpers.

Frontend JS/TS Runtime Modules

Module Files Lines Branches Funcs Missing branches Risk Primary gap
src/pages/explorer 31 97.81 93.43 97.95 95 P0 Browse/search route state, infinite pages, annotations, paper view branches.
src/components/explorer-paper 35 99.00 94.53 98.73 40 P1 Paper contact sheet/detail/search result render branches and empty/error states.
src/app 15 99.87 97.50 100.00 13 P1 Router/shell fallback and runtime state branches.
src/components/shell 7 99.50 94.50 100.00 12 P1 Navigation history, search palette, status/topbar fallbacks.
src/pages/dashboard 10 100.00 94.48 100.00 10 P1 Dashboard fallback/dashboard paper branches have line coverage but weak branch pins.
src/lib/explorer-preferences.ts 1 94.59 83.33 100.00 4 P1 Persistent explorer view-mode parse and localStorage failure paths.
src/lib/paper-preferences.ts 1 100.00 80.00 100.00 4 P1 Appearance preference parse/localStorage fallbacks.
src/components/intelligence 19 100.00 99.07 100.00 5 P1 Browsing rhythm optional/empty branches.
src/components/sidebar 3 100.00 97.32 100.00 3 P1 Background status variant branches.
src/pages/assistant 6 99.30 98.00 98.08 3 P1 Provider-gated assistant page fallback and action branches.
src/pages/import 8 99.76 99.31 99.09 3 P0 Import workflow edge branches and failure UX.
src/components/primitives 9 100.00 98.41 100.00 2 P2 Background progress display branches.
src/components/heatmap 2 100.00 95.56 100.00 2 P2 Calendar/heatmap presentational boundary branches.
src/pages/audit 6 99.67 99.71 98.85 1 P1 Audit page fallback function branch.
src/pages/intelligence 34 100.00 99.89 100.00 1 P1 Paper intelligence panel optional branch.
src/pages/settings 32 100.00 99.89 100.00 1 P1 Data migration section branch.
src/lib/backend-preview-support.ts 1 96.15 98.39 100.00 1 P1 Browser-preview support failure branch.
src/lib/backend-preview-shell-commands.ts 1 100.00 98.91 100.00 1 P1 Shell command fixture branch.
src/lib/backend.ts 1 99.02 100.00 98.72 0 P1 Legacy/browser-preview function residual. Do not expand surface.
src/main.tsx 1 100.00 100.00 100.00 0 P0 Desktop entry is covered and mutation-gated by desktop-contract slice.
src/lib/backend-client 15 100.00 100.00 100.00 0 P0 IPC client surface covered; keep mutation focus on contracts.
src/lib/ipc 4 100.00 100.00 100.00 0 P0 Desktop bridge covered and mutation-gated by desktop-contract slice.
src/lib/core-intelligence 10 100.00 100.00 100.00 0 P1 Deterministic API/client covered; later mutation still needed.
src/lib/i18n 29 100.00 100.00 100.00 0 P1 Catalog parity covered by separate i18n gate.
src/components/review 10 100.00 100.00 100.00 0 P1 PME review primitives covered.
src/pages/schedule 4 100.00 100.00 100.00 0 P0 UI covered; Rust scheduler remains the higher-risk owner.
src/pages/security 4 100.00 100.00 100.00 0 P0 UI covered; Rust app-lock/keyring remain higher-risk owners.
Other single-purpose UI/helper modules 18 100.00 100.00 100.00 0 P2 Keep as regression surface; deprioritize until P0/P1 gaps close.

Top JS branch residual files:

File Branches Missing branches Why it matters
src/pages/explorer/index.tsx 90.35 22 Main Browse/Search route orchestrator.
src/pages/explorer/hooks/use-explorer-infinite-pages.ts 77.66 21 Large-archive infinite scroll and prefetch behavior.
src/components/explorer-paper/paper-contact-sheet.tsx 87.27 14 Browse rendering, virtualization entry point.
src/app/shell.tsx 83.82 11 Global shell runtime/fallback behavior.
src/pages/explorer/paper-view.tsx 86.57 9 Paper Browse/Search composition.
src/pages/dashboard/index.tsx 72.41 8 Dashboard fallback branches.
src/components/explorer-paper/paper-day-insights-helpers.ts 89.74 8 Day aggregate presentation parity.
src/components/explorer-paper/paper-detail-panel.tsx 93.33 7 Notes/tags/detail interaction states.

Rust Modules

The current Rust verifier reports 100% semantic source-line and function coverage across the full src-tauri/**/src/*.rs workspace. Remaining Rust gaps are therefore behavioral, branch, concurrency, I/O, and mutation gaps.

Module Risk Primary gap to audit next
vault-core/archive P0 Ingest/write rollback, dedup edge cases, history query semantics, og-image fetch I/O failures.
vault-core/takeout P0 Malformed/partial payloads, zip read failures, source evidence limits.
browser-history-parser/{chromium,firefox,safari,takeout} P0 Real-browser schema drift, missing optional tables, malformed DBs, timestamp boundaries.
browser-history-fixtures P0 Sidecar table fixture gaps already tracked in BACKLOG.
vault-core/migration P0 Partial writes, manifest tamper, wrong key, forward schema migration, rollback recoverability.
vault-core/app_lock and vault-platform/keyring P0 Auth-disabled states, host keyring failures, session/key recovery boundaries.
vault-platform/scheduler P0 Host command failures, permission/mismatch branches, cross-platform semantics.
pathkeep-desktop/dev_ipc_bridge and worker_bridge P0 Desktop command truth, payload validation, poisoned state, CORS/localhost boundary.
vault-worker/archive_flows P0 Background job cancellation, queue concurrency, og-image worker throttling/retry behavior.
vault-core/intelligence* and vault-worker/intelligence* P1 Optional AI/provider gating, stale sidecars, queue replay/cancel, deterministic fallback parity.
vault-core/annotations P1 URL notes/tags limits, search dimensions, FTS behavior.
vault-core/visit_taxonomy P1 Regional taxonomy, URL normalization, CJK/script-aware tokenization.

Ordered Work Queue

  1. [x] src/pages/explorer/hooks/use-explorer-infinite-pages.ts
    • Reason: P0 large-archive performance and correctness owner; 21 missing JS branches; recent BROWSE-VIRT changes make this a mutation-priority module.
    • Behavior focus: no duplicate in-flight fetches, no prefetch past pageCount, cache-token reset, silent background prefetch failure, load guards.
  2. [x] src/pages/explorer/index.tsx
    • Main route orchestrator; many branches likely need integration-level route tests instead of shallow render assertions.
  3. [x] src/components/explorer-paper/paper-contact-sheet.tsx
    • Virtualization/render-state branches; must assert mounted/recycled/session behavior rather than DOM snapshots alone.
  4. [x] src/components/explorer-paper/paper-detail-panel.tsx
    • Notes/tags persistence UX; assert disabled/error/loading states and mutation-prone handler behavior.
  5. [x] src/lib/explorer-preferences.ts and src/lib/paper-preferences.ts
    • Persistent local preference parse/fallback failure paths.
  6. [x] src/app/shell.tsx and src/components/shell/*
    • Route history and global shell state branches.
  7. [x] src/pages/dashboard/index.tsx
    • Branch coverage is low despite line coverage; assert fallback semantics.
  8. [x] Rust import/fixture sidecar backlog blocks
    • Follow existing BACKLOG order: sidecar fixture extension, minor integrity pins, parser ordering, concurrency.
  9. [x] Rust migration/security/scheduler fault-injection sweep
    • Add tests for partial I/O, permission errors, command failures, and concurrent state transitions before full mutation.
    • [x] 9A migration import fail-fast refusal paths.
    • [x] 9B security/keyring refusal and recovery paths.
    • [x] 9C scheduler host-command failure paths.
  10. [x] JS coverage residual restoration
    • Restored coverage:js from the temporary 99/98/99/99 floor back to the documented 100/100/100/100 gate without excluding active runtime code.
    • [x] 10A Search/Browse chip and result residuals.
    • [x] 10B Explorer route and paper-view residuals.
    • [x] 10C Dashboard/app shell residuals.
    • [x] 10D Hook/helper branch residuals.
    • [x] 10E Explorer runtime hooks and paper helper residuals.
    • [x] 10F Explorer Paper component residuals.
    • [x] 10G Shared component, preview facade, and route residuals.
  11. [x] Rust parser mutation hardening
    • Used focused cargo-mutants evidence to compensate for the unavailable stable Rust branch metric without claiming branch coverage that the tool does not report.
    • Hardened browser-history-parser Chromium/Firefox/Safari/Takeout streaming, schema observation, evidence chunk, source sniffing, native key, warning, and merge-report contracts until the parser mutation slice reached 0 missed mutants.

Bug / Drift Register

ID Status Evidence Action
QA-GAP-001 closed JS coverage command previously passed at 99/98/99/99 while quality matrix says 100/100/100/100. Module 10G restored vitest.config.ts thresholds to 100/100/100/100 after bun run coverage:js reported All files 100/100/100/100 and lcov had no uncovered JS lines, branches, or functions. Resolved in Module 10G; keep 100/100/100/100 as the per-commit JS runtime gate.
QA-GAP-002 closed Rust coverage verifier has no branch metric, and cargo llvm-cov --branch --manifest-path src-tauri/Cargo.toml --workspace --all-features --lcov --output-path coverage/rust-branch.lcov.info fails on the pinned stable Rust 1.94.1 toolchain because -Z coverage-options=branch is nightly-only. Before hardening, bun run mutation:rust:parser reported 431 mutants with 78 missed, 298 caught, and 55 unviable. After focused parser tests, the same parser slice reported 431 mutants with 376 caught, 55 unviable, and 0 missed. Resolved by mutation-resistant parser behavior tests. Do not claim Rust branch coverage from llvm-cov; keep Rust branch confidence tied to focused boundary/error tests plus cargo mutants evidence under the stable toolchain.
QA-GAP-003 closed src/pages/explorer/index.tsx still built a paginated PaperExplorerView prop, but the current render grammar routes every infiniteDisabled condition to Search, grouped views, invalid-regex callout, or locked/uninitialized states before PaperExplorerView can mount. Resolved in Module 10B by deleting the route-only paginated prop, the redundant results render guard, and the unreachable infinite-scroll fallback. Component-level pagination remains covered in paper-view.test.tsx.
QA-GAP-004 closed src/components/explorer-paper/paper-contact-sheet.tsx retained defensive guards that targeted behavior could not naturally hit: toolbar ref missing after mount and canLoadMore false after the sentinel exists. Resolved in Module 10F by simplifying the redundant guards and keeping contact-sheet behavior pinned through component and virtualization assertions.
QA-GAP-005 closed src/components/explorer-paper/paper-detail-panel.tsx retained unreachable defensive branches: layout-flush with pendingFlushRef but no active timer, and LookFurtherRow non-interactive rendering even though the parent filters out rows without handlers. Resolved in Module 10F by simplifying the redundant branches and keeping visible detail-panel behavior pinned through focused component assertions.
QA-GAP-006 closed Shell residual branches were defensive or redundant under current public grammar: shellStorage() without window, handleSearchQuery() receiving a blank query after PKSearchPalette already suppresses blank searches, archiveHealthy ?? false after a boolean expression, domainAbbreviation().split('.')[0] ?? cleaned, and duplicate route-history location keys inside an effect keyed by location.key. Resolved in Module 10C by deleting redundant shell/palette fallbacks, pinning palette whitespace suppression, and covering the route-history duplicate-key branch through a StrictMode render.

Checkpoint Log

Baseline

  • JS: bun run coverage:js passed, 2023 tests, global 99 statements / 98 branches / 99.51 functions / 99.5 lines.
  • Rust: bun run coverage:rust passed, verifier 100% across 35246 semantic source lines and 1634 source functions.
  • No product bugs confirmed yet; only gate drift recorded.

Module 1: src/pages/explorer/hooks/use-explorer-infinite-pages.ts

Added 5 focused behavior tests:

  • cache-token refresh clears accumulated page 2..N state even when the query signature is unchanged.
  • a second loadMore() while page 2 is in-flight does not issue a duplicate foreground request.
  • unconditional background prefetch does not request pageCount + 1.
  • background prefetch rejection stays silent when the foreground page succeeds.
  • downward +2 prefetch does not request beyond the reported pageCount.

Commands:

bunx vitest run src/pages/explorer/hooks/use-explorer-infinite-pages.test.tsx
bunx vitest run src/pages/explorer/hooks/use-explorer-infinite-pages.test.tsx --coverage --coverage.include=src/pages/explorer/hooks/use-explorer-infinite-pages.ts --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Test Files  1 passed (1)
Tests       15 passed (15)

use-explorer-infinite-pages.ts targeted coverage:
statements 92.02 | branches 79.78 | functions 95.65 | lines 99.01
uncovered line: 345

Movement from repository baseline for this file:

  • statements: 91.30 -> 92.02
  • branches: 77.66 -> 79.78
  • functions: 95.65 -> 95.65
  • lines: 99.01 -> 99.01

Suspected bugs: none. The uncovered residual is mostly the hard MAX_ACCUMULATED_PAGES cap path, which is awkward to hit without either exporting a test seam or driving hundreds of loadMore() transitions. Keep it queued for a dedicated cap-boundary test rather than adding an artificial coverage-only assertion.

Module 2: src/pages/explorer/index.tsx

Added 4 route-shell behavior tests and tightened the paper-surface mocks:

  • Paper filter strip apply trims filled values, deletes blank values, clears stale page, removes individual chips through updateParam, and delegates clear-all to the URL-state hook.
  • Legacy config.ogImage.fetchEnabled = false is folded into fetchMode: 'off' before the route calls useExplorerOgImages.
  • Search-result URLs such as Google SERPs suppress misleading og:image hydration even when the row/cache advertises an image.
  • Detail panel "All of domain" resets the URL to a domain-only Browse state, dropping search/date/profile/pagination context.

Commands:

bunx vitest run src/pages/explorer/index.test.tsx
bunx vitest run src/pages/explorer/index.test.tsx --coverage --coverage.include=src/pages/explorer/index.tsx --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0
bun run typecheck

Actual output:

Test Files  1 passed (1)
Tests       17 passed (17)

index.tsx targeted coverage from this test file:
statements 86.54 | branches 79.82 | functions 65.3 | lines 86.25
uncovered lines include 606, 659-698, 867-933

Suspected product bugs: none confirmed. Drift recorded as QA-GAP-003: the paginated PaperExplorerView prop branch appears unreachable under the current route grammar, so it should be resolved as product-code cleanup or restored as an intentional mode rather than covered with artificial tests.

Module 3: src/components/explorer-paper/paper-contact-sheet.tsx

Added a focused behavior test file with 10 assertions over the contact-sheet branches most likely to survive mutation:

  • Cards-mode frame numbers continue across day boundaries, so virtualization extraction cannot reset the filmstrip counter per day.
  • Sticky day headers use the measured toolbar height both with and without ResizeObserver, including when filter chips wrap the toolbar.
  • Invalid visit timestamps render --:--; malformed/unrepresentable day keys render raw labels instead of crashing.
  • Null page titles fall back to sanitized URL text in both cards and list mode.
  • Target clear is safe when the caller omits onClearTarget.
  • Infinite-scroll cap guidance, non-intersecting sentinel, and load-error alert states render with the intended copy and side effects.

Commands:

bunx vitest run src/components/explorer-paper/paper-contact-sheet.behavior.test.tsx
bunx vitest run src/components/explorer-paper/paper-contact-sheet.test.tsx src/components/explorer-paper/paper-contact-sheet.behavior.test.tsx src/components/explorer-paper/paper-contact-sheet.virt.test.tsx src/components/explorer-paper/paper-contact-sheet.spike.test.tsx --coverage --coverage.include=src/components/explorer-paper/paper-contact-sheet.tsx --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Behavior test file:
Test Files  1 passed (1)
Tests       10 passed (10)

All paper-contact-sheet focused files:
Test Files  4 passed (4)
Tests       45 passed (45)

paper-contact-sheet.tsx targeted coverage:
statements 98.14 | branches 98.18 | functions 100 | lines 100
uncovered lines: 279, 677

Full checkpoint gate:
bun run check
JS coverage: statements 99.27 | branches 98.24 | functions 99.66 | lines 99.69
Rust coverage: 100% for 35246 instrumented source lines and 1634 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Suspected product bugs: none confirmed. Drift recorded as QA-GAP-004: the remaining guards are defensive branches that current render grammar does not make observable through user behavior.

Module 4: src/components/explorer-paper/paper-detail-panel.tsx

Added 5 behavior assertions around the detail panel's persistence and navigation surface:

  • External notes refreshes update the textarea when no local edit is pending.
  • Pending local notes survive a stale backend refresh and still debounce-save the local draft.
  • Look-further rows are suppressed when route handlers are not wired, avoiding phantom navigation labels.
  • Look-further rows remain clickable when count hints are omitted.
  • Favicon and og:image media use default test ids when the caller does not pass a panel testId.

Commands:

bunx vitest run src/components/explorer-paper/paper-detail-panel.test.tsx --coverage --coverage.include=src/components/explorer-paper/paper-detail-panel.tsx --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Test Files  1 passed (1)
Tests       28 passed (28)

paper-detail-panel.tsx targeted coverage:
statements 100 | branches 96.19 | functions 100 | lines 100
uncovered lines: 228, 757-765

Full checkpoint gate:
bun run check
JS coverage: statements 99.27 | branches 98.27 | functions 99.66 | lines 99.69
Rust coverage: 100% for 35246 instrumented source lines and 1634 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Suspected product bugs: none confirmed. Drift recorded as QA-GAP-005: the remaining branches are defensive paths that the current public render grammar does not expose through user behavior.

Module 5: src/lib/explorer-preferences.ts and src/lib/paper-preferences.ts

Added 9 behavior assertions over persistent local preference fallbacks:

  • Explorer view mode and clock format return defaults and no-op safely when window is unavailable.
  • Explorer read helpers return defaults when localStorage read access throws.
  • Re-persisting the current clock format skips storage writes and does not emit a redundant live-update event.
  • Paper preferences return defaults and no-op safely when window is unavailable.
  • Unrecognized paper preference values normalize back to the shipped appearance.
  • applyPaperPreferences returns the supplied candidate without touching globals when both window and document are unavailable.
  • applyPaperPreferences still updates document attributes and dispatches the live-update event when localStorage persistence fails.

Commands:

bunx vitest run src/lib/explorer-preferences.test.ts src/lib/paper-preferences.test.ts --coverage --coverage.include=src/lib/explorer-preferences.ts --coverage.include=src/lib/paper-preferences.ts --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Test Files  2 passed (2)
Tests       37 passed (37)

explorer-preferences.ts targeted coverage:
statements 100 | branches 100 | functions 100 | lines 100

paper-preferences.ts targeted coverage:
statements 100 | branches 100 | functions 100 | lines 100

Full checkpoint gate:
bun run check
JS coverage: statements 99.34 | branches 98.35 | functions 99.66 | lines 99.71
Rust coverage: 100% for 35246 instrumented source lines and 1634 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Suspected product bugs: none confirmed.

Module 11: Rust parser mutation hardening

Added or strengthened parser behavior tests around the Rust branch/mutation gap that cannot be measured by stable cargo llvm-cov --branch:

  • Chromium and Firefox incremental URL queries now prove that a single zero cursor does not accidentally take the first-import fast path.
  • Chromium/Firefox/Safari streaming tests now assert exact batch boundaries, retained vs streamed evidence movement, and capability populated/total row counts.
  • Shared schema observation tests pin required/missing table rows, optional table status, row counts, primary-key ordinals, and NOT NULL detection.
  • Source-evidence chunk tests pin the empty contract across search, navigation, engagement, context, and native entity families.
  • Takeout source tests now pin direct localized history recognition, whitespace normalization, non-Chrome JSON no-sniff behavior, sniff byte limits, and the "Browser History" + "time_usec" conjunction.
  • Takeout payload tests now pin native primary key fallback/source fields, missing-time warning absence/presence, callback-abort display text, canonical-only adapter passthrough, merge-report count/capability/evidence aggregation, and default optional HistoryBatchConsumer no-op methods.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p browser-history-parser
bun run mutation:rust:quality
bun run coverage:rust
bun run check

Actual output:

Focused parser tests:
browser-history-parser: 56 passed; 0 failed

Rust branch metric probe:
cargo llvm-cov --branch ... failed on pinned stable Rust 1.94.1 because
`-Z coverage-options=branch` is nightly-only.

Rust mutation quality:
bun run mutation:rust:quality
browser-history-parser: 431 mutants tested in 11m; 376 caught; 55 unviable; 0 missed
vault-core ai helper filter: Found 0 mutants to test

Rust coverage:
Rust coverage verified at 100% for 35643 instrumented source lines and
1654 source functions.

Full checkpoint gate:
bun run check
format/lint/i18n/typecheck: passed
unit: 279 files passed; 2162 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
Rust workspace tests: browser-history-parser 56 passed; vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
JS coverage: All files statements 100%, branches 100%, functions 100%, lines 100%
Rust coverage: verified at 100% for 35643 instrumented source lines and 1654 source functions
supply-chain/platform/release checks: passed
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed. Drift closed as QA-GAP-002; Rust branch coverage is still not claimed from llvm-cov under the stable toolchain, and branch confidence for this slice is now backed by focused boundary/error tests plus zero missed parser mutants.

Module 8A: WORK-IMPORT-FIXTURE-SIDECARS-A

Added 6 focused Rust behavior tests:

  • Chromium fixture self-validation proves generated downloads, keyword search terms, favicon bitmap bytes, and icon mappings round-trip through the production parser.
  • Chromium Favicons fixture writer overwrites an existing invalid file with a queryable companion schema and does not retain duplicate rows on rewrite.
  • T6 asserts Chromium downloads rows land in archive downloads with source id, paths, byte counts, state, MIME fields, and Unix-ms start time preserved.
  • T7 asserts keyword_search_terms rows land in archive search_terms linked to the canonical URL with term text, normalized term, profile id, and keyword id preserved.
  • T8 asserts favicon page URLs match canonical URL rows and identical synthetic PNG payload bytes deduplicate into one favicon_blobs row.
  • T9 asserts multiple icon_mapping rows for one icon create separate page URL favicon rows while preserving the shared icon URL.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p browser-history-fixtures write_favicons_overwrites_existing_file_with_companion_schema
cargo test --manifest-path src-tauri/Cargo.toml -p browser-history-fixtures chromium --tests
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core chromium_sidecars --lib

Actual output:

browser-history-fixtures:
write_favicons_overwrites_existing_file_with_companion_schema: 1 passed
chromium_roundtrip: 4 passed

vault-core:
Test result: ok. 4 passed; 0 failed; 651 filtered out.

Full checkpoint gate:

bun run check
JS unit: Test Files 277 passed (277), Tests 2076 passed (2076)
JS coverage: statements 99.37 | branches 98.59 | functions 99.66 | lines 99.72
Rust coverage: 100% for 35596 instrumented source lines and 1643 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Suspected product bugs: none confirmed.

Module 8B: WORK-IMPORT-TEST-MINOR-A

Added 5 focused Rust behavior tests:

  • E10 asserts Chromium URL visit_count and typed_count values round-trip for both zero-count typed URLs and nonzero visited URLs.
  • E11 asserts a dangling Chromium from_visit reference is preserved verbatim instead of being rewritten to NULL or 0.
  • E12 asserts Chromium visits.visit_duration lands unchanged in the current archive visits.visit_duration_ms column.
  • E13 asserts Safari history_visits.synthesized persists to the cold source-evidence DB as safari.synthesized with source ids and source_field intact.
  • E14 asserts Firefox moz_historyvisits.visit_type values land in visits.transition_type without Chromium normalization.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-core e1 --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core --lib

Actual output:

e1 filter: 6 passed; 0 failed; 654 filtered out (existing E1 + new E10-E14)
vault-core lib: 660 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

Full checkpoint gate:

bun run check
JS unit: Test Files 277 passed (277), Tests 2076 passed (2076)
JS coverage: statements 99.38 | branches 98.61 | functions 99.66 | lines 99.72
Rust coverage: 100% for 35805 instrumented source lines and 1648 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Suspected product bugs: none confirmed.

Module 8C: WORK-IMPORT-TEST-PARSER-ORDERING-A

Added 1 focused Rust behavior test:

  • chunk_consumer_skips_visits_when_url_batch_has_not_populated_the_map asserts the current ArchiveChunkConsumer::visits contract when a parser emits a visit before its URL batch: the visit is skipped silently, no canonical visit row is inserted, skipped progress increments, imported/duplicate progress stays at zero, and last_visit_id is not advanced.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-core chunk_consumer_skips_visits_when_url_batch_has_not_populated_the_map --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core --lib

Actual output:

parser-ordering targeted: 1 passed; 0 failed; 660 filtered out
vault-core lib: 661 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

Full checkpoint gate:

bun run check
JS unit: Test Files 277 passed (277), Tests 2076 passed (2076)
JS coverage: statements 99.37 | branches 98.59 | functions 99.66 | lines 99.72
Rust coverage: 100% for 35835 instrumented source lines and 1649 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Gate note: the first full bun run check attempt hit a non-reproducible existing settings-shell-b navigation timing failure under coverage:js. The targeted test passed without and with coverage, standalone coverage:js passed, and the subsequent full bun run check passed.

Suspected product bugs: none confirmed.

Module 8D: WORK-IMPORT-TEST-CONCURRENCY-A

Added 1 focused Rust behavior test:

  • same_profile_writer_waits_for_committed_watermark asserts same-profile concurrent archive writers are serialized at the SQLite transaction boundary: the second writer cannot read profile_watermarks while the first writer's transaction is uncommitted, and it observes the committed cursor after the first writer commits.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-core same_profile_writer_waits_for_committed_watermark --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core --lib

Actual output:

concurrency targeted: 1 passed; 0 failed; 661 filtered out
vault-core lib: 662 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

Full checkpoint gate:

bun run check
JS unit: 277 files passed; 2076 tests passed
JS desktop contract: 5 files passed; 26 tests passed; coverage 100% statements/branches/functions/lines
JS coverage: statements 99.38%; branches 98.61%; functions 99.66%; lines 99.72%
Rust workspace tests: vault-core 662 passed in base; vault-core 663 passed under coverage cfg
Rust coverage: 100% for 35835 instrumented source lines and 1649 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 score; 64 mutants; 0 survived; 0 timed out

Suspected product bugs: none confirmed. The audit found no separate app-level ingest queue; the current same-profile guarantee is SQLite writer-lock serialization after upsert_source_profile, documented in import-dedup-audit.md §4.1.

Module 8E: WORK-MAINT-IMPORT-EDGE-CASES-SPLIT-A

Maintained the import edge-case scenario safety net while reducing the oversized owner file:

  • dedup_scenarios_edge_cases.rs now holds only shared fixture/env helpers and child module declarations.
  • Test bodies moved into focused owners: chromium_contracts, empty_and_resilience, time_and_nullable, unicode_and_flags, and minor_data_integrity.
  • All 19 existing edge-case test names and assertions were preserved.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-core archive::ingest::dedup_scenarios_edge_cases --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core --lib

Actual output:

edge-case targeted: 19 passed; 0 failed; 643 filtered out
vault-core lib: 662 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

Full checkpoint gate:

bun run check
JS unit: 277 files passed; 2076 tests passed
JS desktop contract: 5 files passed; 26 tests passed; coverage 100% statements/branches/functions/lines
JS coverage: statements 99.37%; branches 98.59%; functions 99.66%; lines 99.72%
Rust workspace tests: vault-core 662 passed in base; vault-core 663 passed under coverage cfg
Rust coverage: 100% for 35835 instrumented source lines and 1660 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 score; 64 mutants; 0 survived; 0 timed out

Suspected product bugs: none confirmed. This was a maintainability-only split; audit §4 / §6 links now point at the focused owner modules.

Module 8F: WORK-MAINT-IMPORT-INGEST-FACADE-SPLIT-A

Maintained the ingest orchestrator safety net while reducing the oversized facade:

  • ingest/mod.rs now contains the production facade only: profile selection, stream dispatch, chunk consumer, watermark advancement, and source-evidence plan persistence.
  • The embedded low-level regression suite moved to ingest/core_tests.rs.
  • A responsibility map in import-dedup-audit.md records why production ArchiveChunkConsumer and source-evidence persistence stayed together for now.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-core archive::ingest::core_tests --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core --lib

Actual output:

core_tests targeted: 7 passed; 0 failed; 655 filtered out
vault-core lib: 662 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

Full checkpoint gate:

bun run check
unit: 277 files passed; 2076 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.38%, branches 98.61%, functions 99.66%, lines 99.72%
Rust workspace tests: vault-core 662 passed; vault-platform 46 passed; vault-worker 70 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed. This was a maintainability-only split; test assertions and product ingest semantics were preserved.

Module 9A: vault-core::migration import refusal fault paths

Added two behavior assertions for whole-app import refusal ordering:

  • Wrong encrypted source key must fail with the typed invalid-key prefix while preserving the existing target archive, derived marker, and all non-.bak paths.
  • Payload hash mismatch must fail before preservation/install renames, leaving the existing target archive and derived marker untouched and creating no .bak-* sidecars.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-core migration::fault_tests --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core --lib

Actual output:

migration::fault_tests: 2 passed; 0 failed; 662 filtered out
vault-core lib: 664 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

Full checkpoint gate:

bun run check
unit: 277 files passed; 2076 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.38%, branches 98.61%, functions 99.66%, lines 99.72%
Rust workspace tests: vault-core 664 passed; vault-platform 46 passed; vault-worker 70 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed. These tests pin the current intended fail-fast behavior before the live project tree is renamed or overwritten.

Module 9B: vault-core::app_lock and vault-platform::keyring

Added two behavior assertions for security refusal and recovery paths:

  • Malformed App Lock session state fails closed for status and unlock, malformed secret hydration fails, and clearing the passcode removes the corrupted state and secret files so the user can recover.
  • Provider API keys do not satisfy database-key saved status, and clearing a provider key does not remove the database key.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-core malformed_app_lock_files_fail_closed_until_passcode_is_cleared --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-platform provider_key_does_not_satisfy_database_key_status_or_clear_database_secret --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-core app_lock --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-platform keyring --lib

Actual output:

app_lock targeted: 1 passed; 0 failed; 664 filtered out
keyring targeted: 1 passed; 0 failed; 46 filtered out
app_lock module: 5 passed; 0 failed; 660 filtered out
keyring module: 4 passed; 0 failed; 43 filtered out

Full checkpoint gate:

bun run check
unit: 277 files passed; 2076 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.38%, branches 98.61%, functions 99.66%, lines 99.72%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed. These tests pin fail-closed behavior for corrupted lock files while keeping the documented recovery path available, and they prevent provider API key storage from being mistaken for database key storage.

Module 9C: vault-platform::scheduler

Strengthened one host-command failure behavior:

  • macOS launchctl bootstrap failure must leave apply_schedule non-applied, preserve the generated plist, write an audit artifact with the bootstrap status, return an error verification check, and make the next schedule_status report permission-warning instead of installed.

Commands:

cargo test --manifest-path src-tauri/Cargo.toml -p vault-platform macos_apply_schedule_reports_bootstrap_failures_without_erroring --lib
cargo test --manifest-path src-tauri/Cargo.toml -p vault-platform scheduler --lib

Actual output:

macOS bootstrap-failure targeted: 1 passed; 0 failed; 46 filtered out
scheduler module: 35 passed; 0 failed; 12 filtered out

Full checkpoint gate:

bun run check
unit: 277 files passed; 2076 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.38%, branches 98.61%, functions 99.66%, lines 99.72%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed. The scheduler audit found existing Windows/macOS host-command failure coverage for access denied, missing tasks, mismatches, loaded-without-file states, and manual Linux paths; this checkpoint tightens the macOS failed-bootstrap recovery/status contract.

Module 10A: JS Search/Browse chip and result residuals

Added focused frontend behavior assertions and removed two dead defensive guards:

  • Advanced search help now has direct coverage for its default test id plus the keyboard focus/blur popover path.
  • Search-filter helpers now pin escaped quotes inside quoted operators, smart-quoted operands, one-character operands, and quoted-empty operands that must not surface as chips.
  • Search result rows now assert unrelated keys do not select the row, missing transition labels stay hidden, and title fallback proceeds from title to URL to domain.
  • Paper search panel now has a child-contract test proving stale remove ids are ignored and focus scheduling tolerates a child that does not attach the input ref.
  • Deleted the unreachable highlightQuery regex try/catch and the unreachable appendFilterOperator unchanged-query guard. Query tokens are escaped before regex construction, and the panel only appends valid hard-coded tag / note operators.

Commands:

bunx vitest run src/components/explorer-paper/paper-advanced-search-help.test.tsx src/components/explorer-paper/paper-search-hero.test.tsx src/pages/explorer/paper-search-filters.test.ts src/components/explorer-paper/paper-search-result.test.tsx src/pages/explorer/paper-search-panel.test.tsx src/pages/explorer/paper-search-panel.child-contract.test.tsx --coverage --coverage.include=src/components/explorer-paper/paper-advanced-search-help.tsx --coverage.include=src/pages/explorer/paper-search-filters.ts --coverage.include=src/components/explorer-paper/paper-search-result.tsx --coverage.include=src/pages/explorer/paper-search-panel.tsx --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Test Files 6 passed (6)
Tests 60 passed (60)
Targeted coverage for paper-advanced-search-help.tsx, paper-search-filters.ts,
paper-search-result.tsx, and paper-search-panel.tsx:
statements 100 | branches 100 | functions 100 | lines 100

Full checkpoint gate:

bun run check
unit: 279 files passed; 2085 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.48%, branches 98.75%, functions 99.72%, lines 99.77%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Gate note: the first full bun run check attempt in the command sandbox failed only in mockito-backed Rust tests because local test servers could not bind sockets (Operation not permitted). Re-running the same command with normal local permissions passed.

Suspected product bugs: none confirmed.

Module 10B: JS Explorer route and paper-view residuals

Added 10 focused frontend behavior assertions and removed route/helper branches that the current public grammar cannot exercise:

  • PaperExplorerView now pins calendar day selection, non-Escape key handling, invalid active-date labels, omitted todayIso current-day fallback, malformed clock-format events, locale-format fallback/no-crash behavior, and decorated pagination / infinite-scroll descriptors.
  • PaperDetailPanelMount now asserts "All of domain" forwards the selected domain and closes the panel.
  • ExplorerPage now asserts enabled og:image settings forward their configured fetch mode and desktop command transport selects the desktop annotation store.
  • Deleted the route-only paginated PaperExplorerView prop, the redundant || results render guard, and the unreachable infiniteDisabled ? undefined branch at the Browse route mount point. The route grammar already sends every infiniteDisabled surface to Search, grouped views, invalid-regex callouts, or locked/uninitialized states before Browse mounts.
  • Simplified Paper search day-group sorting and Paper view density override helpers to remove equality / Map.has branches that were unreachable under the helper contracts.

Commands:

bunx vitest run src/pages/explorer/index.test.tsx src/pages/explorer/paper-view.test.tsx src/pages/explorer/paper-detail-panel-mount.test.tsx src/pages/explorer/paper-view-helpers.test.ts src/pages/explorer/paper-search-helpers.test.ts
bunx tsc -b --pretty false
bun run coverage:js

Actual output:

Targeted tests:
Test Files 5 passed (5)
Tests 94 passed (94)

coverage:js:
Test Files 279 passed (279)
Tests 2095 passed (2095)
All files statements 99.6 | branches 99 | functions 99.84 | lines 99.88

Module 10B files now at 100/100/100/100:
src/pages/explorer/index.tsx
src/pages/explorer/paper-detail-panel-mount.tsx
src/pages/explorer/paper-search-helpers.ts
src/pages/explorer/paper-view-helpers.ts
src/pages/explorer/paper-view.tsx

Full checkpoint gate:

bun run check
format/lint/i18n/typecheck: passed
unit: 279 files passed; 2095 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.61%, branches 99.02%, functions 99.84%, lines 99.88%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed.

Module 10C: JS Dashboard/app shell residuals

Added 6 focused frontend behavior assertions and removed public-grammar duplicate branches:

  • Exported and tested the route registry lookup guard so unknown route handles fail loudly instead of silently picking the wrong shell screen.
  • Pinned epigraph storage values that omit the date/index separator so daily rotation overwrites malformed local state.
  • Pinned PKSearchPalette whitespace-only queries so the palette stays on the empty branch and does not call the backend search path.
  • Pinned useRouteHistoryNav under StrictMode so duplicate effect mounts do not move the shell history stack twice for the same location key.
  • Pinned Dashboard On This Day target-date formatting for non-English resolved locales.
  • Pinned Dashboard archive-access fallback cleanup so a queued non-probe clear cannot update state after unmount.
  • Deleted redundant shellStorage() SSR fallback, the shell-level blank search guard that PKSearchPalette already makes unreachable, the redundant archiveHealthy ?? false fallback, and the unreachable domainAbbreviation().split('.')[0] ?? cleaned fallback.

Commands:

bunx vitest run src/app/index-tests/router-structure.test.tsx src/app/shell-helpers.test.ts src/app/shell.test.tsx src/components/shell/pk-search-palette.test.tsx src/components/shell/use-route-history-nav.test.tsx src/pages/dashboard/on-this-day-card.test.tsx src/pages/dashboard/route-fallback.test.tsx
bunx tsc -b --pretty false
bunx vitest run src/app/index-tests/router-structure.test.tsx src/app/shell-helpers.test.ts src/app/shell.test.tsx src/components/shell/pk-search-palette.test.tsx src/components/shell/use-route-history-nav.test.tsx src/pages/dashboard/on-this-day-card.test.tsx src/pages/dashboard/route-fallback.test.tsx --coverage --coverage.include=src/app/router.tsx --coverage.include=src/app/shell-helpers.ts --coverage.include=src/app/shell.tsx --coverage.include=src/components/shell/pk-search-palette.tsx --coverage.include=src/components/shell/use-route-history-nav.ts --coverage.include=src/pages/dashboard/on-this-day-card.tsx --coverage.include=src/pages/dashboard/route-fallback-access.ts --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0
bun run coverage:js

Actual output:

Targeted tests:
Test Files 7 passed (7)
Tests 109 passed (109)

Targeted 10C coverage:
All files 100/100/100/100
src/app/router.tsx 100/100/100/100
src/app/shell-helpers.ts 100/100/100/100
src/app/shell.tsx 100/100/100/100
src/components/shell/pk-search-palette.tsx 100/100/100/100
src/components/shell/use-route-history-nav.ts 100/100/100/100
src/pages/dashboard/on-this-day-card.tsx 100/100/100/100
src/pages/dashboard/route-fallback-access.ts 100/100/100/100

coverage:js:
Test Files 279 passed (279)
Tests 2101 passed (2101)
All files statements 99.63 | branches 99.09 | functions 99.84 | lines 99.89

Full coverage groups now at 100/100/100/100:
src/app
src/components/shell
src/pages/dashboard

Full checkpoint gate:

bun run check
format/lint/i18n/typecheck: passed
unit: 279 files passed; 2101 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.64%, branches 99.11%, functions 99.84%, lines 99.89%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed.

Module 10D: JS Explorer hook/helper residuals

Added 15 focused hook/helper behavior assertions and removed duplicate private guards that the public hook grammar could not exercise:

  • url-state-derivations now pins keyword grouped views, positive page parsing, empty query/cursor history payloads, null query signatures, semantic/session active-filter chips, bare recent-search labels, and incomplete date shortcut ranges.
  • useDesktopAnnotations now treats missing backend tags as an empty list during hydration instead of leaking undefined into the UI state.
  • useBrowseDayInsightsCache now pins stale success/failure replies after a refresh-key rotation and removed a private duplicate request guard already made unreachable by resolve()'s cache-entry check.
  • useExplorerArchiveDensity now pins stale success/failure replies, cancelled not-ready reset microtasks, available-year-only bounds, and no-finite-year responses. The removed sequence ref duplicated effect cleanup cancellation.
  • useExplorerData now pins cached adjacent prefetch reuse, stale multi-page prefetch cancellation, cancellation before the transition paint, and empty semantic recall results without inventing a selected row.

Commands:

bunx vitest run src/pages/explorer/url-state-derivations.test.ts src/pages/explorer/use-desktop-annotations.test.tsx src/pages/explorer/hooks/use-browse-day-insights-cache.test.tsx src/pages/explorer/hooks/use-explorer-archive-density.test.tsx src/pages/explorer/hooks/use-explorer-data.test.tsx
bunx vitest run src/pages/explorer/url-state-derivations.test.ts src/pages/explorer/use-desktop-annotations.test.tsx src/pages/explorer/hooks/use-browse-day-insights-cache.test.tsx src/pages/explorer/hooks/use-explorer-archive-density.test.tsx src/pages/explorer/hooks/use-explorer-data.test.tsx --coverage --coverage.include=src/pages/explorer/url-state-derivations.ts --coverage.include=src/pages/explorer/use-desktop-annotations.ts --coverage.include=src/pages/explorer/hooks/use-browse-day-insights-cache.ts --coverage.include=src/pages/explorer/hooks/use-explorer-archive-density.ts --coverage.include=src/pages/explorer/hooks/use-explorer-data.ts --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Targeted tests:
Test Files 5 passed (5)
Tests 58 passed (58)

Targeted 10D coverage:
All files 100/100/100/100
src/pages/explorer/url-state-derivations.ts 100/100/100/100
src/pages/explorer/use-desktop-annotations.ts 100/100/100/100
src/pages/explorer/hooks/use-browse-day-insights-cache.ts 100/100/100/100
src/pages/explorer/hooks/use-explorer-archive-density.ts 100/100/100/100
src/pages/explorer/hooks/use-explorer-data.ts 100/100/100/100

Full checkpoint gate:

bun run check
format/lint/i18n/typecheck: passed
unit: 279 files passed; 2114 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.69%, branches 99.21%, functions 99.84%, lines 99.91%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed.

Module 10E: JS Explorer runtime hook and paper helper residuals

Added 24 focused behavior assertions and removed redundant branches inside Explorer runtime hooks/helpers:

  • useExplorerInfinitePages now pins disabled/no-head dormancy, same-key buffer retention, cancelled rejected foreground pages, directional prefetch failure, directional prefetch reuse, and the hard-cap/page-count decision through a pure derivation helper. The hook now treats pageCount as the numbered-page contract instead of mixing in cursor-style hasNext.
  • useExplorerOgImages now pins empty result windows, cache-token reset of the pending mark-shown batch, and unmount timer cleanup. Redundant enqueued-url tracking was removed because the visible URL dedupe, inflight set, and cache map already prevent duplicate local reads/refetch enqueues.
  • useScrollDirection now pins stable same-direction samples, RAF dedupe, and pending-frame cleanup.
  • useViewportMount now pins no-ref setup, empty observer callbacks, zero-height recycle, and re-entry measurement behavior.
  • groupEntriesByDay/format helpers now pin out-of-range date fallback and hour12 formatting. Private empty-session and manual date-comparator branches were simplified under the public grouping contract.
  • ExplorerDetailPanel now pins loading, empty, missing metadata, null-domain, and null-profile detail-rail branches.

Commands:

bunx vitest run src/pages/explorer/hooks/use-explorer-infinite-pages.test.tsx src/pages/explorer/hooks/use-explorer-og-images.test.tsx src/pages/explorer/hooks/use-scroll-direction.test.tsx src/pages/explorer/hooks/use-viewport-mount.test.tsx src/pages/explorer/paper/group-entries.test.ts src/pages/explorer/panels/detail-panel.test.tsx
bunx vitest run src/pages/explorer/hooks/use-explorer-infinite-pages.test.tsx src/pages/explorer/hooks/use-explorer-og-images.test.tsx src/pages/explorer/hooks/use-scroll-direction.test.tsx src/pages/explorer/hooks/use-viewport-mount.test.tsx src/pages/explorer/paper/group-entries.test.ts src/pages/explorer/panels/detail-panel.test.tsx --coverage --coverage.include=src/pages/explorer/hooks/use-explorer-infinite-pages.ts --coverage.include=src/pages/explorer/hooks/use-explorer-og-images.ts --coverage.include=src/pages/explorer/hooks/use-scroll-direction.ts --coverage.include=src/pages/explorer/hooks/use-viewport-mount.ts --coverage.include=src/pages/explorer/paper/group-entries.ts --coverage.include=src/pages/explorer/panels/detail-panel.tsx --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Targeted tests:
Test Files 6 passed (6)
Tests 81 passed (81)

Targeted 10E coverage:
All files 100/100/100/100
src/pages/explorer/hooks/use-explorer-infinite-pages.ts 100/100/100/100
src/pages/explorer/hooks/use-explorer-og-images.ts 100/100/100/100
src/pages/explorer/hooks/use-scroll-direction.ts 100/100/100/100
src/pages/explorer/hooks/use-viewport-mount.ts 100/100/100/100
src/pages/explorer/paper/group-entries.ts 100/100/100/100
src/pages/explorer/panels/detail-panel.tsx 100/100/100/100

Full checkpoint gate:

bun run check
format/lint/i18n/typecheck: passed
unit: 279 files passed; 2138 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.87%, branches 99.56%, functions 99.87%, lines 99.92%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed.

Module 10F: JS Explorer Paper component residuals

Added 11 focused behavior assertions and removed redundant private branches in Explorer Paper presentation/runtime components:

  • PaperCalendarPopover now pins low- and medium-density hover spark opacity instead of relying on the year picker to incidentally walk density tiers.
  • aggregateDayInsights now pins empty URL handling, empty domain+URL fallback, invalid timestamp handling, exact-hour durations, and the Intl.NumberFormat unit-format fallback.
  • PaperDayInsights now pins a details disclosure with no peak-hour row, host time-formatting failures, and long revisited-URL truncation.
  • PaperListRow now pins og:image as the list icon fallback when favicon bytes are absent.
  • Redundant branches were simplified where the public render grammar already guarantees the ref or handler: assistant auto-scroll, contact-sheet toolbar measurement, infinite-scroll footer canLoadMore, detail-panel pending flush and look-further rows, filter-strip outside-click container, and top-domain max-visit lookup.

Commands:

bunx vitest run src/components/explorer-paper/paper-assistant-view.test.tsx src/components/explorer-paper/paper-calendar-popover.test.tsx src/components/explorer-paper/paper-contact-sheet.test.tsx src/components/explorer-paper/paper-contact-sheet.behavior.test.tsx src/components/explorer-paper/paper-contact-sheet.virt.test.tsx src/components/explorer-paper/paper-day-insights-helpers.test.ts src/components/explorer-paper/paper-day-insights.test.tsx src/components/explorer-paper/paper-detail-panel.test.tsx src/components/explorer-paper/paper-filter-strip.test.tsx src/components/explorer-paper/paper-browse-primitives.test.tsx
bunx vitest run src/components/explorer-paper/paper-assistant-view.test.tsx src/components/explorer-paper/paper-calendar-popover.test.tsx src/components/explorer-paper/paper-contact-sheet.test.tsx src/components/explorer-paper/paper-contact-sheet.behavior.test.tsx src/components/explorer-paper/paper-contact-sheet.virt.test.tsx src/components/explorer-paper/paper-day-insights-helpers.test.ts src/components/explorer-paper/paper-day-insights.test.tsx src/components/explorer-paper/paper-detail-panel.test.tsx src/components/explorer-paper/paper-filter-strip.test.tsx src/components/explorer-paper/paper-browse-primitives.test.tsx --coverage --coverage.include=src/components/explorer-paper/paper-assistant-view.tsx --coverage.include=src/components/explorer-paper/paper-calendar-popover.tsx --coverage.include=src/components/explorer-paper/paper-contact-sheet.tsx --coverage.include=src/components/explorer-paper/paper-day-insights-helpers.ts --coverage.include=src/components/explorer-paper/paper-day-insights.tsx --coverage.include=src/components/explorer-paper/paper-detail-panel.tsx --coverage.include=src/components/explorer-paper/paper-filter-strip.tsx --coverage.include=src/components/explorer-paper/paper-list-row.tsx --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Targeted tests:
Test Files 10 passed (10)
Tests 173 passed (173)

Targeted 10F coverage:
All files 100/100/100/100
src/components/explorer-paper/paper-assistant-view.tsx 100/100/100/100
src/components/explorer-paper/paper-calendar-popover.tsx 100/100/100/100
src/components/explorer-paper/paper-contact-sheet.tsx 100/100/100/100
src/components/explorer-paper/paper-day-insights-helpers.ts 100/100/100/100
src/components/explorer-paper/paper-day-insights.tsx 100/100/100/100
src/components/explorer-paper/paper-detail-panel.tsx 100/100/100/100
src/components/explorer-paper/paper-filter-strip.tsx 100/100/100/100
src/components/explorer-paper/paper-list-row.tsx 100/100/100/100

Full checkpoint gate:

bun run check
format/lint/i18n/typecheck: passed
unit: 279 files passed; 2149 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 99.94%, branches 99.76%, functions 99.87%, lines 99.95%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed.

Module 10G: JS shared component, preview facade, and route residuals

Added or extended 14 focused behavior assertions and removed two redundant private branches in the final JS coverage residual set:

  • YearHeatmap click behavior now depends on a single observable clickable predicate, while covered tests still pin count-bearing date selection.
  • Browsing-rhythm calendar/card tests now pin disabled year navigation, missing reset shortcuts, ready-state summaries, zero-visit cells, and selection-prompt behavior.
  • Background progress/status tests now pin empty labels, log-only detail rows, AI queue fallback totals, and warning precedence.
  • Browser-preview facade tests now pin og:image default hydration, dashboard earliest/latest ordering stability, and the legacy unsupported prefetch_og_images rejection instead of expanding src/lib/backend.ts.
  • Paper assistant/audit/import route tests now pin paper-layout send behavior, providerless fallback attribution, manifest-chain selection, and browser/file method normalization.
  • Intelligence and settings residuals now pin title fallback rendering and the data-migration apply path without retaining an unreachable phase guard.

Commands:

bunx vitest run src/components/heatmap/year-heatmap.test.tsx src/components/intelligence/browsing-rhythm-calendar.test.tsx src/components/intelligence/browsing-rhythm-card.test.tsx src/components/primitives/background-progress.test.tsx src/components/sidebar/background-status.test.tsx src/lib/backend-preview-shell-commands.test.ts src/lib/backend-preview-showcase.test.ts src/lib/backend-tests/preview-workflows.test.ts src/pages/intelligence-surfaces/assistant-and-shell.test.tsx src/pages/audit/index.test.tsx src/pages/import/index.test.tsx src/pages/intelligence/paper-intelligence-panel.test.tsx src/pages/settings/data-migration-section.test.tsx
bunx vitest run src/lib/backend-preview-showcase.test.ts src/pages/import/index.test.tsx
bun run coverage:js

Actual output:

Focused 10G sweep:
Test Files 13 passed (13)
Tests 103 passed (103)

Final residual recheck:
Test Files 2 passed (2)
Tests 17 passed (17)

JS coverage:
Test Files 279 passed (279)
Tests 2162 passed (2162)
All files statements 100%, branches 100%, functions 100%, lines 100%
No uncovered JS lines, branches, or functions in lcov.info.

Full checkpoint gate:

bun run check
format/lint/i18n/typecheck: passed
unit: 279 files passed; 2162 tests passed
desktop contract: 5 files passed; 26 tests passed; coverage 100/100/100/100
JS coverage: All files statements 100%, branches 100%, functions 100%, lines 100%
Rust workspace tests: vault-core 665 passed; vault-platform 47 passed; vault-worker 70 passed
Rust coverage cfg tests: vault-core 666 passed; vault-platform 49 passed; vault-worker 80 passed
Rust coverage: verified at 100% for 35459 instrumented source lines and 1652 source functions
supply-chain/platform/release checks: passed
build: passed
browser E2E: 4 passed
desktop bridge E2E: 3 passed
desktop contract mutation: 64 mutants; 100.00 score; 0 survived; 0 timed out

Suspected product bugs: none confirmed. Drift closed as QA-GAP-001; the JS runtime coverage floor is now the documented 100/100/100/100 gate.

Module 6: src/app/shell.tsx and src/components/shell/*

Added 15 behavior assertions across the global shell and shell chrome:

  • Background busy payloads render the non-blocking progress strip and do not show the blocking overlay.
  • Malformed preference-change events leave the current theme chrome unchanged.
  • Dashboard totals, last-archive telemetry, and source color fallback order are visible in the status bar.
  • Backend palette errors and missing items payloads render the no-results state instead of silently passing.
  • URL-only, titleless, URL-less, and no-visit-date palette hits map to visible, selectable results and route to Explorer as expected.
  • Escape actually closes the palette; Manage Sources actually routes to Settings; stale palette searches do not leak late success or failure results.
  • Status bar epigraph/profile-label fallbacks, topbar navigator fallback, route replace navigation, blank-UA shortcut labels, and isContentEditable shortcut suppression are pinned with observable assertions.

Commands:

bunx vitest run src/app/shell.test.tsx src/components/shell/pk-search-palette.test.tsx src/components/shell/pk-status-bar.test.tsx src/components/shell/pk-topbar.test.tsx src/components/shell/use-route-history-nav.test.tsx --coverage --coverage.include=src/app/shell.tsx --coverage.include=src/components/shell/pk-search-palette.tsx --coverage.include=src/components/shell/pk-status-bar.tsx --coverage.include=src/components/shell/pk-topbar.tsx --coverage.include=src/components/shell/use-route-history-nav.ts --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Test Files  5 passed (5)
Tests       68 passed (68)

Targeted shell coverage:
All files: statements 99.2 | branches 97.79 | functions 100 | lines 100
shell.tsx: statements 98.8 | branches 95.58 | functions 100 | lines 100
pk-search-palette.tsx: statements 100 | branches 97.14 | functions 100 | lines 100
pk-status-bar.tsx: statements 100 | branches 100 | functions 100 | lines 100
pk-topbar.tsx: statements 100 | branches 100 | functions 100 | lines 100
use-route-history-nav.ts: statements 98.83 | branches 98.27 | functions 100 | lines 100

Full checkpoint gate:

bun run check
JS unit: Test Files 277 passed (277), Tests 2070 passed (2070)
JS coverage: statements 99.36 | branches 98.51 | functions 99.66 | lines 99.72
Rust coverage: 100% for 35246 instrumented source lines and 1634 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Suspected product bugs: none confirmed. Drift recorded as QA-GAP-006 for the remaining defensive/redundant branches that are not naturally reachable through the current public shell grammar.

Module 7: src/pages/dashboard/index.tsx

Added 6 focused behavior tests:

  • archive span uses the current day when latestVisitAt is missing, instead of rendering the missing-span placeholder.
  • read-model derived archive state renders zero-size fallback, encrypted mode, source count, database-path fallback, and missing manifest hash.
  • getOnThisDay() null data renders the empty state instead of preserving stale entries.
  • stale successful and failed On This Day responses are discarded after the route becomes uninitialized.
  • On This Day entries/header actions, year heatmap actions, and active-thread actions navigate to their route-level destinations.
  • morning greeting branch is covered alongside the existing afternoon/evening branches.

Commands:

bunx vitest run src/pages/dashboard/index.test.tsx --coverage --coverage.include=src/pages/dashboard/index.tsx --coverage.thresholds.lines=0 --coverage.thresholds.branches=0 --coverage.thresholds.functions=0 --coverage.thresholds.statements=0

Actual output:

Test Files  1 passed (1)
Tests       17 passed (17)

Targeted dashboard coverage:
All files: statements 100 | branches 100 | functions 100 | lines 100
index.tsx: statements 100 | branches 100 | functions 100 | lines 100

Full checkpoint gate:

bun run check
JS unit: Test Files 277 passed (277), Tests 2076 passed (2076)
JS coverage: statements 99.38 | branches 98.61 | functions 99.66 | lines 99.72
Rust coverage: 100% for 35246 instrumented source lines and 1634 source functions
Browser E2E: 4 passed
Desktop bridge E2E: 3 passed
Desktop contract mutation: 100.00 mutation score, 64 mutants, 0 survived

Suspected product bugs: none confirmed.