feat: lazy-load memory browse and Flashbacks

[nonnil]

Summary:

• Implemented Task 20 lazy-loading performance workflow end to end.
• Replaced initial /memories archive-wide loading with cursor-paginated SQLite browse pages.
• Kept search, read-state, taxonomy, and ?flashback=<id> filters server-global before pagination.
• Split Flashback loading into bounded right-rail recent rows, memory-card batches for visible memory IDs, and reader All-tab lazy loading.
• Scoped add-memory success revalidation to first browse page + taxonomy before reader navigation.
• Preserved backup failsafe behavior and fixed one backup test harness edge where Git hook env variables broke nested temp-repo git commands.

Subtask / commit mapping:

• 20.1 Browse query and page contract: d9d8100
• 20.2 SQLite repository pagination: 28def27
• 20.3 Browse loader contract: 3c5a24b
• 20.4 Memories infinite scroll UI: fd43d41
• 20.5 Lazy Flashback browse data: 4babf16
• 20.6 Reader lazy Flashback tab: 00dc970
• 20.7 Integration verification fixes: 6a1d3f2, eddf0df

Behavior covered:

• /memories initial route loads a 30-row first page.
• Infinite scroll / Load more appends the next page without duplicate titles.
• /memories?q=<term> searches the full archive, including rows not loaded on the first page.
• /memories?flashback=<id> still finds the owning memory and hydrates the selected excerpt.
• Right-rail recent Flashbacks use loadRecentFlashbackBrowseRows({ limit: 5 }), not all memories.
• Reader Current tab no longer starts the global Flashback query.
• Reader All tab starts the global Flashback query on demand.
• Add-memory success no longer revalidates the old all-memory browse query.
• Backup failsafe warning/alert behavior remains covered by the backup drift/inconsistency tests.

Verification:

• mise exec -- bun run test tests/memories/browse-data.test.ts tests/components/browse-data-query.test.ts: passed, 2 files / 28 tests.
• mise exec -- bun run test tests/server/db/schema.test.ts tests/server/db/repositories.test.ts: passed, 2 files / 36 tests.
• mise exec -- bun run test tests/server/memories/browse.test.ts tests/server/browse-loaders.test.ts tests/server/flashbacks/repository.test.ts: passed, 3 files / 24 tests.
• mise exec -- bun run test tests/components/browse-loader.test.ts tests/components/flashbacks-loader.test.ts tests/components/app-shell.test.ts tests/components/memory-browse-actions.test.ts tests/components/reader-flashback-tabs.test.ts: passed, 5 files / 62 tests.
• mise exec -- bun run test tests/components/reader-flashback-tabs.test.ts tests/components/reader-memory-loader.test.ts tests/components/memory-reader-actions.test.ts tests/components/memory-reader-flashback-selection.test.ts: passed, 4 files / 36 tests.
• mise exec -- bun run test tests/server/backup/backup-environment.test.ts with parent GIT_DIR/GIT_WORK_TREE set: passed, 1 file / 10 tests.
• mise exec -- bun run typecheck: passed.
• mise exec -- bun run test: passed, 103 files / 799 passed / 5 todo.
• mise exec -- bun run build: passed. Existing Node DEP0155 warning remains from defuddle -> temml.
• mise exec -- bun run test:e2e: passed, 47 Playwright tests.
• Pre-push hook on final push: passed typecheck, full test, and build.

Performance evidence:

Measured with the Task 20 harness against a copy of the local /Users/vvx/.trauma dataset to avoid mutating the live database/store. Dataset: 110 memories, first page 30 rows, 39 renderable Flashbacks, largest source memory 249,025 bytes.

Path	Before	After median	Result
Initial browse memory load	loadBrowseMemories() 721 ms, 110 rows, 39 nested Flashbacks	loadBrowseMemoryPage() 1.28 ms	30 rows, nextCursor=true
Memory-card Flashbacks	Included in all-memory browse payload	16.59 ms	30 memory IDs, 5 card Flashbacks
Right-rail recent Flashbacks	Derived from all-memory browse payload / loadFlashbackBrowseRows() 756 ms	15.96 ms	5 recent Flashbacks
Large reader source	322 ms first / 256 ms warm	227.67 ms	Same 249 KB source memory
Reader Current tab	Could coexist with global Flashback resource creation	~0 ms shortcut mapping	1 current-memory shortcut, no global query
Reader All tab	Global query could start during initial reader render	587.24 ms, deferred	39 Flashbacks, only after All
Add-memory success + reader/page refresh	Revalidated all-memory browse path	3.71 ms with synthetic importer	new reader ready, first page remains 30 rows

Known local state:

• The pushed branch tip is eddf0df.
• The linked worktree currently has unrelated uncommitted changes that appeared during final verification, including Task 21 workflow/docs/UI/settings changes and Task 20 workflow deletions. They were not staged or committed for this PR.
• .sawyer/exclude-whitelist.txt was checked; whitelisted paths such as .tmp/, TASK_19_INSTRUCTION.md, TEMP_MAIN_CONTENT_EXTRACTION_VALIDATION_REPORT.md, app.config.timestamp_1778391094301.js, and refined_sample/ were not staged.

Backup safety:

• Backup failsafe test coverage still emits and asserts drift/inconsistency alerts.
• The branch does not change backup integrity algorithms or weaken failsafe checks.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: ff0b6831-60af-4612-a485-58d5aeaf815d

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/lazy-loading

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[nonnil]

review @copilot @codex

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: eddf0df15b

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".

[nonnil]

review @copilot @codex

Addressed the three Codex review threads in 7a8a4bf and pushed the branch. Threads were replied to and resolved; local verification and the pre-push hook passed.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Keep them coming!

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".