feat(server): sync thread read status across clients

[t3dotgg]

Thread read and unread state lived in web local storage, so browsers and devices could disagree about which completed threads had been seen.

This moves lastViewedAt into the event-sourced server projection and adds server-owned mark-viewed and mark-unread commands. Web and mobile now update the shared state, older servers are capability-gated, and view updates do not touch updatedAt or reorder threads. Existing threads remain read by default after migration.

Verified with focused decider, projection, persistence, command, reducer, sidebar, and UI-state tests (177 passing), plus scoped typechecks for contracts, client-runtime, server, web, and mobile.

Built with GPT-5.6-sol via the Codex harness in T3 Code.

---

Note

Medium Risk
Touches orchestration commands, DB migration backfill, and cross-client read semantics; migration timestamps could briefly skew unread indicators until users view threads again.

Overview
Thread read/unread is now server-owned instead of per-browser threadLastVisitedAtById in uiStateStore. Threads carry lastViewedAt through contracts, projection (projection_threads + migration 039), snapshot queries, and client reducers; view updates emit thread.view-status-updated without bumping updatedAt.

New orchestration commands thread.mark-viewed and thread.mark-unread (with optimistic concurrency on viewed boundaries) are wired through client-runtime atoms (markViewed / markUnread) and advertised via the threadViewStatus capability on the server descriptor.

Web and mobile call markThreadViewed when opening a thread (and on wake/settle flows where applicable); sidebars, wake banners, and context menus use lastViewedAt and hasUnseenWake, and hide mark-unread when the server lacks support. Existing threads are backfilled on migrate so upgrades do not flood unread state.

^{Reviewed by Cursor Bugbot for commit 0125c6f. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Sync thread read/unread status across clients via server-owned view state

• Replaces client-local threadLastVisitedAtById timestamps in uiStateStore with server-persisted lastViewedAt, tracked via new thread.mark-viewed and thread.mark-unread orchestration commands.
• Adds threadViewStatus capability to the server descriptor; clients check this capability before issuing view-state mutations, so older servers are unaffected.
• The decider handles thread.mark-viewed (with optimistic concurrency via expectedLastViewedAt/supersededViewedAt) and thread.mark-unread (sets lastViewedAt to 1ms before the latest turn completion), emitting thread.view-status-updated events.
• Migration 039 adds last_viewed_at to projection_threads and backfills existing rows with the migration timestamp.
• Sidebar unread indicators, wake banners, and context-menu 'Mark unread' items now derive state from thread.lastViewedAt and are hidden when the server lacks threadViewStatus support.
• Risk: existing threads are backfilled with the migration time as last_viewed_at, so threads that were previously read may briefly reappear as unread after the migration runs.

^{Macroscope summarized 0125c6f.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. 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: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 99cc9244-8c4b-4e3b-8b93-4634f836c57d

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

---

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.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	11.3 KiB	11.3 KiB	+9 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	5.5 KiB	5.5 KiB	+11 B (+0.2%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	5.9 KiB	5.9 KiB	−2 B (−0.0%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	49.7 KiB	49.7 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	16	16	0 (0.0%)	21	✅
Claude	Total thread wire	11.3 KiB	11.4 KiB	+18 B (+0.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	5.5 KiB	5.5 KiB	+11 B (+0.2%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	5.9 KiB	5.9 KiB	+7 B (+0.1%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	50.6 KiB	50.6 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	16	16	0 (0.0%)	21	✅

Baseline: 0640410 · PR result: 0125c6f · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 94.7 KiB
• Claude decoded thread snapshot: 95.4 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR adds a new feature for syncing thread read/unread status across clients via the server. It includes a database migration, new orchestration commands and events, and changes how read state is tracked (from client-local to server-synchronized). The scope and nature of these changes warrant human review.

No code changes detected at 0125c6f. Prior analysis still applies.

^{You can customize Macroscope's approvability policy. Learn more.}

[t3dotgg]

Terrible when compared to #5652

[t3dotgg]

Wrong pr my bad, too many tabs lol

[macroscopeapp]

🟡 Medium src/orchestration.ts:674

ThreadMarkViewedCommand.viewedAt is documented as a server-owned boundary copied from thread completion/wake state, but the schema accepts any client-supplied IsoDateTime. Because the decider persists this value directly (only comparing it against the previous one), a client can submit a far-future timestamp and make all subsequent completions and wakes appear already read. The server should derive or validate this boundary against actual thread state rather than trusting the client value.

🚀 Reply "fix it for me" or copy this AI Prompt for your agent:

In file @packages/contracts/src/orchestration.ts around line 674:

`ThreadMarkViewedCommand.viewedAt` is documented as a server-owned boundary copied from thread completion/wake state, but the schema accepts any client-supplied `IsoDateTime`. Because the decider persists this value directly (only comparing it against the previous one), a client can submit a far-future timestamp and make all subsequent completions and wakes appear already read. The server should derive or validate this boundary against actual thread state rather than trusting the client value.

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 0125c6f. Configure here.}

[cursor]

Stale view token rejects open-after-unread

Medium Severity

markViewed snapshots expectedLastViewedAt from the shell before the per-thread serial queue runs. If mark-unread is ahead in that queue or has landed on the server before the shell refreshes, the following open-thread mark-viewed still sends the pre-unread token, fails the decider invariant, and leaves the thread unread while the user is viewing it.

Additional Locations (1)

• apps/server/src/orchestration/decider.ts#L451-L459

 

^{Reviewed by Cursor Bugbot for commit 0125c6f. Configure here.}