[teamleaderleo]

Leo, 2026-11-01, on an inbox message line in the transcript: "i wanted this to be left aligned with the other tool calls". On the folded message group ("2 messages from 1 session") under a stack of tool rows: "let's just make this look better".

What changed

Every folded group of messages from other sessions now has a tool-row head. On base these groups are drawn by three components:

• the waiting group, PendingMessagesCard ("N messages from M sessions", the row Leo named);
• a delivered inbox batch, LaneNoticeBatch ("N inbox threads");
• the legacy InboxNoticeCard group.

Each had a bare chevron as its leading glyph, a 26px line, and words 1.5px right of the tool rows'. All three now share MessageGroupHead.tsx/.css, which is a tool row (.cv-tool):

• Icon: the sender's harness mark in the tool rows' icon column. With several senders it shows the messages glyph (agent.chat). There is no leading chevron.
• Size: the tool rows' 22px height, icon-to-word gap and type size.
• Text: the count reads as the summary, with the session names muted after it (hidden in a narrow pane, as before).
• Chevron: after the label. It shows and lights only on hover, focus or while the group is open, as an opened tool group's steps do.
• What opening shows is unchanged, and so are the existing toggle class names (cv-pending__toggle, cv-lane-batch__toggle). A report batch ("N reviews ended") keeps its own head. The row estimate (model.ts) follows the 22px heads.

A sender's mark is its real harness, never a guess from its name. SenderMark (from IncomingMessageCard.tsx, shared by received lines, inbox lines and the group heads) takes the listed chat's harness, else the fromHarness acpmux bound. With neither, it draws the neutral agent glyph. Before, it fell back to a brand looked up from the sender's name, so an unlisted Codex lane called claude-… drew Claude's mark. The notice's fromHarness now reaches the delivered batch's received lines (laneReport.ts, laneNoticeBatches.ts). The root cause is that acpmux outbox sends bind no fromSessionId/fromHarness. That fix belongs in acpmux and is filed as #30247.

Single inbox line. On base its glyph center and title x already equal the tool rows' (#29124 folded #28796's alignment). The new browser test pins that, so it stays aligned among tool rows.

Not changed: tool rows themselves, the status rows between replies (#29124) and subagent rows (#29276). Each row's content, colours and glass are kept.

Gallery

agent-pane.inbox-notice has new versions. Each shows, among tool runs: a received line, an unlisted claude-… sender (neutral glyph), a delivered one-sender group, a delivered several-sender group, and the waiting "2 messages from 1 session".

• icon-column: shipped. The chevron shows on hover.
• icon-column-look-chevron: the chevron always shown.
• icon-column-look-count: the count alone, with senders in the tooltip.
• icon-column-open: the several-sender group opened.

The existing look-lines-* and look-card-* versions stay beside them. I picked icon-column: it matches the tool rows exactly, and the always-on chevron repeats a control on every row.

Tests

• webviews/test/inbox-row-icon-column.test.ts (Chromium, light and dark): every received line's and group head's glyph center and title x equal a tool row's (0.5px). Each group head's height, icon-to-word gap and type size equal a tool row's, and its chevron is hidden at rest and shown on hover.
  • Red with the base PendingMessagesCard/LaneNoticeBatch at the same gallery: the delivered batches' glyph is at 32px vs the tool rows' 28, their title at 46.5px vs 42.5, and they are 26px tall vs 22. The waiting group's title is at 44px vs 42.5, its gap 8px vs 6.5, and it is 26px tall.
  • Green after: every group is glyph 28px, text 42.5px, gap 6.5px, 22px tall and 14px type, the same as the tool rows.
• webviews/test/inbox-sender-mark.test.tsx: a Claude sender (listed, or bound by fromHarness) gets the Claude mark. A listed Codex chat named claude-sr2-fleet-mesh-2 gets OpenAI's. An unknown claude-x sender gets the neutral glyph. A one-sender group gets its mark and a mixed group the messages glyph. Red on base (3 of 5 failed), green after.
• Still green: transcript-icon-column, incoming-sender-chip, lane-notice-card, incoming-message-disclosure-browser, agent-pane-suggested-tasks, InboxNoticeCard.test.tsx, check:pieces and tsc. All ran on the fleet through cmux-ci test.

Spec: design/chat-transcript.md "One icon column". The inbox-row rule is added in a leo-workspace PR that cites this one.

Risk: low. The change is limited to the three message-group heads' markup and CSS, the row estimate for those heads, and the sender-mark fallback. A sender whose name alone used to pick a brand now shows the neutral glyph until acpmux binds its harness (#30247).

🤖 Generated with Claude Code

Agent session: agent session

[teamleaderleo]

Owned by an agent session · session

[cubic-dev-ai]

2 issues found across 9 files

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="webviews/test/inbox-sender-mark.test.tsx">

<violation number="1" location="webviews/test/inbox-sender-mark.test.tsx:140">
P2: This mixed-sender assertion does not verify the messages glyph: `marks()` classifies every SVG as `"neutral"`, so an incorrect icon would still pass. Assert the icon’s `data-icon` value directly.</violation>
</file>

<file name="webviews/src/agent-session/acpmux/model.ts">

<violation number="1" location="webviews/src/agent-session/acpmux/model.ts:1659">
P2: Changing the inbox look does not invalidate the virtual layout, so this grouped predicate is not applied to existing offscreen rows immediately. Switching between a folded look and `lines`/`card` can leave stale heights and cause virtualization to mount or position rows at the wrong offsets; make the layout subscribe to the inbox-look change or pass a look version into its memo dependencies.</violation>
</file>

<sub>Reply with feedback, questions, or to request a fix.

View guided diff | Turn on auto-fix | Re-trigger cubic</sub>

[cubic-dev-ai]

P2: This mixed-sender assertion does not verify the messages glyph: marks() classifies every SVG as "neutral", so an incorrect icon would still pass. Assert the icon’s data-icon value directly.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At webviews/test/inbox-sender-mark.test.tsx, line 140:

<comment>This mixed-sender assertion does not verify the messages glyph: `marks()` classifies every SVG as `"neutral"`, so an incorrect icon would still pass. Assert the icon’s `data-icon` value directly.</comment>

<file context>
@@ -0,0 +1,141 @@
+  await render({
+    threads: [thread("grid-lane", { fromSession: "s-claude" }), thread("claude-x", { threadId: "pr:3" })],
+  });
+  expect(marks(".cv-inbox__head > .cv-tool__icon")).toEqual(["neutral"]);
+});
</file context>

[cubic-dev-ai]

P2: Changing the inbox look does not invalidate the virtual layout, so this grouped predicate is not applied to existing offscreen rows immediately. Switching between a folded look and lines/card can leave stale heights and cause virtualization to mount or position rows at the wrong offsets; make the layout subscribe to the inbox-look change or pass a look version into its memo dependencies.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At webviews/src/agent-session/acpmux/model.ts, line 1659:

<comment>Changing the inbox look does not invalidate the virtual layout, so this grouped predicate is not applied to existing offscreen rows immediately. Switching between a folded look and `lines`/`card` can leave stale heights and cause virtualization to mount or position rows at the wrong offsets; make the layout subscribe to the inbox-look change or pass a look version into its memo dependencies.</comment>

<file context>
@@ -1654,11 +1654,13 @@ function measuredRowHeight(
-    if (grouped && !messageOpened(sessionId, row.id)) return finish(line + MESSAGE_GAP);
+    // The group's head is a tool row (22px) in 2px margins (InboxNoticeCard.css .cv-inbox--group).
+    const head = 2 + 22 + 2;
+    const grouped = threads.length > 1 && foldsThreads(inboxNoticeLook());
+    if (grouped && !messageOpened(sessionId, row.id)) return finish(head + MESSAGE_GAP);
     const { lineHeight } = metrics();
</file context>

[teamleaderleo]

Review (codex, router-review): SHIP at ea98d18

Reviewer: codex gpt-6.1-sol xhigh, fast tier.

Follow-ups: 1. Each lands as an issue in this repo that links this receipt; the lane that opened the PR fixes it forward.

SHIP at ea98d18

Visual verdict: no images; Leo tries the build

1. [P2][follow-up] webviews/src/agent-session/acpmux/conversation/InboxNoticeCard.tsx:221 reverses the thread list before selecting the single-sender group’s lead. With multiple threads from one sender carrying different harness/session metadata, this selects the oldest matching thread and can display a stale mark. Select the newest matching thread instead.

Follow-ups:

• [P2][follow-up] Fix single-sender group mark selection to use the newest matching thread.

[cmux-bot]

cmux-bot approves ea98d18: an independent review passed this exact head (router-review receipt by teamleaderleo). A push needs a new review; a later BLOCK at this head dismisses this approval.

[github-actions]

cmux status _{(one comment, edited in place; each line is updated by the job that owns it)}

Review 661da86: HQ feed card · gallery before/after runs · checks

App of this head (app.zip, kept 1 day)

gh workflow run reload-build.yml -R manaflow-ai/cmux --ref feat-cmux-next -f tag=pr30652 -f ref=inbox-row-align-harness-icon -f platform=macos -f backend_mode=local

Gallery before/after for 661da86: 16 states rendered, 1 images differ, 2814 more touched states not rendered. Images are in the run's artifact, not on this public PR; the independent review of this head reads them on its own, so the PR needs no hand-made screenshots of these states.

gh run download 41315828921 -R manaflow-ai/cmux -n gallery-advisory-30652-661da862005b3d4a926b97fc6481fd7bbd845556 (then read gallery-review.json)

States

State	Theme	Status	Diff	Picked by
agent-pane.addressed-message/relayed	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.addressed-message/relayed	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/prompt-rest	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/prompt-rest	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/prompt-hover	Apple System Colors	changed	3.88%	own IncomingMessageCard.tsx
agent-pane.card-copy/prompt-hover	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/prompt-copied	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/prompt-copied	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/message-rest	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/message-rest	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/message-hover	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/message-hover	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/message-copied	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/message-copied	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/quiet-rest	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/quiet-rest	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/quiet-copied	Apple System Colors Light	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.card-copy/quiet-copied	Apple System Colors	unchanged	0%	own IncomingMessageCard.tsx
agent-pane.chat-width/standard	Apple System Colors Light	unchanged	0%	own App.tsx
agent-pane.chat-width/standard	Apple System Colors	unchanged	0%	own App.tsx
agent-pane.chat-width/wide	Apple System Colors Light	unchanged	0%	own App.tsx
agent-pane.chat-width/wide	Apple System Colors	unchanged	0%	own App.tsx
agent-pane.chat-width/wide-blocks	Apple System Colors Light	unchanged	0%	own App.tsx
agent-pane.chat-width/wide-blocks	Apple System Colors	unchanged	0%	own App.tsx
agent-pane.composer-edge/short-edge	Apple System Colors Light	unchanged	0%	own App.tsx
agent-pane.composer-edge/short-edge	Apple System Colors	unchanged	0%	own App.tsx
agent-pane.composer-edge/wide-edge	Apple System Colors Light	unchanged	0%	own App.tsx
agent-pane.composer-edge/wide-edge	Apple System Colors	unchanged	0%	own App.tsx
agent-pane.composer-edge/no-edge	Apple System Colors Light	unchanged	0%	own App.tsx
agent-pane.composer-edge/no-edge	Apple System Colors	unchanged	0%	own App.tsx
agent-pane.composer-floats/scrolled-to-end	Apple System Colors Light	unchanged	0%	own App.tsx
agent-pane.composer-floats/scrolled-to-end	Apple System Colors	unchanged	0%	own App.tsx

_{Pick other states with a Gallery: entry.id, entry.id/variant line in the PR body.}

Passes: CI passes on 661da86200.

CI passes on 661da86200 (run 41315829010 attempt 1).

Written by scripts/ci/classify_failures.py (ci-failure-attribution.yml); signatures are its SIGNATURES table. A machine verdict is the runner's fault, not this PR's; yours means the failing file is one this PR changes, also red on main that main's latest full suite fails the same way, seen on other PRs that it failed on another pull request's run lately.

red: cmux-next checks (god files, concurrency, crash safety, l10n); no finished base run on c17c8da to compare; do not merge until green; see lane-rules

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.4132201832701 (621470180f)
• cmux NEXT nightly 1.0.0-nightly.4132260480101 (f3ae381789)
• cmux NEXT nightly 1.0.0-nightly.4132310303501 (2ed0f1c3a4)
• cmux NEXT nightly 1.0.0-nightly.4132418488301 (5f712a27cc)
• cmux NEXT nightly 1.0.0-nightly.4132492786501 (3c0c02a5ae)
• cmux NEXT nightly 1.0.0-nightly.4132569907101 (641e45405d)

[blacksmith-sh]

Found 7 test failures on Blacksmith runners:

Failures

Test	View Logs
/tmp/tmp_mwprlhp/summary-fail.json (live):/summary-fail.json (live):	View Logs
/tmp/tmpfocp6w_3/summary-missing.json (live):/summary-missing.json (live):	View Logs
[new_workspaces] debug.new_workspace timed 0 of 50 new workspace rows/ [new_workspaces] debug.new_workspace timed 0 of 50 new workspace rows	View Logs
[new_workspaces] the window draws no frames/[new_workspaces] the window draws no frames	View Logs
[tabs] debug.tab_switch timed 0 of 40 switches/ [tabs] debug.tab_switch timed 0 of 40 switches	View Logs
[tabs] the window draws no frames/[tabs] the window draws no frames	View Logs
test_the_repository_highlights_validate (main.DigestTests.test_the_repository_highl ights_validate)/ test_the_repository_highlights_validate (main.DigestTests.test_the_repository_highl ights_validate)	View Logs

^{Need help on this PR? Tag @codesmith-bot with what you need.}

[teamleaderleo]

Review (codex, router-review): SHIP at a5104be

Reviewer: codex gpt-6-astra high, fast tier.

Follow-ups: 2. Each lands as an issue in this repo that links this receipt; the lane that opened the PR fixes it forward.

SHIP at a5104be

Visual verdict: All 16 available after screenshots match their before images. The changed inbox-group states are absent; Leo tries the build.

1. [P2] [follow-up] webviews/src/agent-session/acpmux/conversation/MessageGroupHead.tsx:23 — groupSender identifies senders only by name. Two distinct sessions named reviewer, one Claude and one Codex, receive the last sender’s harness mark instead of the messages glyph. Confirmed by executing the function’s extracted body. Prefer session IDs when available, consistent with pendingMessages.ts.

2. [P2] [follow-up] review/images/gallery/README.txt:2 — The retained screenshots cover addressed-message and card-copy states, but none show the changed inbox groups. Include the icon-column and opened-group captures to make the alignment and disclosure changes visually reviewable.

No blocking correctness issue found. Full tests were not rerun: Bun and dependencies are unavailable in this box.

Follow-ups:

• Use session identity when selecting a group’s mark; cover duplicate sender names.
• Supply screenshots of the changed inbox-group states.

[cmux-bot]

cmux-bot approves a5104be: an independent review passed this exact head (router-review receipt by teamleaderleo). A push needs a new review; a later BLOCK at this head dismisses this approval.

[teamleaderleo]

Review (codex, router-review): SHIP at 06543de

Reviewer: codex gpt-6-astra high, fast tier.

Follow-ups: 2. Each lands as an issue in this repo that links this receipt; the lane that opened the PR fixes it forward.

SHIP at 06543de

Visual verdict: no images; Leo tries the build

1. [P2][follow-up] [MessageGroupHead.tsx:23] groupSender decides whether a group is mixed using only message.name. Pending groups are actually partitioned by sessionId, so two distinct sessions sharing a display name render one session’s harness mark instead of the messages glyph. Compare session identity when detecting mixed groups.

Follow-ups:

• [P2][follow-up] Fix mixed-group detection to use session identity.
• [follow-up] No after screenshots were provided.

[cmux-bot]

cmux-bot approves 06543de: an independent review passed this exact head (router-review receipt by teamleaderleo). A push needs a new review; a later BLOCK at this head dismisses this approval.

[teamleaderleo]

Review (codex, router-review): SHIP at b5d10b4

Reviewer: codex gpt-6-astra high, fast tier.

Follow-ups: 1. Each lands as an issue in this repo that links this receipt; the lane that opened the PR fixes it forward.

SHIP at b5d10b4

Visual verdict: no images; Leo tries the build

1. [P2][follow-up] groupSender determines whether a group has one sender by comparing display names only (MessageGroupHead.tsx:21). Pending-message grouping is keyed by sessionId when available, and the session store explicitly supports multiple live chats sharing a name. Two sessions with the same name can therefore produce a “2 messages from 2 sessions” group whose head shows one sender’s harness mark instead of the multi-sender agent.chat glyph; a single session whose name changes can have the opposite error. Use the session identity (sessionId ?? name) when choosing the group mark.

Follow-ups:

• [P2][follow-up] Fix group sender detection to use session identity rather than display-name equality.

[cmux-bot]

cmux-bot approves b5d10b4: an independent review passed this exact head (router-review receipt by teamleaderleo). A push needs a new review; a later BLOCK at this head dismisses this approval.

[teamleaderleo]

Review (codex, router-review): SHIP at 661da86

Reviewer: codex gpt-6.1-sol xhigh, fast tier.

Follow-ups: 2. Each lands as an issue in this repo that links this receipt; the lane that opened the PR fixes it forward.

SHIP at 661da86

Visual verdict: no images; Leo tries the build

1. [P2] [follow-up] MessageGroupHead.tsx:23: groupSender identifies senders by name alone. Two sessions named reviewer, with distinct session IDs and harnesses, select one session’s mark instead of the messages glyph. Reproduced directly. Use session identity consistently with bySender.

2. [P2] [follow-up] InboxNotice.gallery.ts:272: No screenshots were supplied for this head, so rendered alignment and hover behavior remain visually unverified.

Syntax parsing and focused harness-propagation checks passed. Bun and browser tests were unavailable because the runtime and dependencies are absent. No blocking findings.

Follow-ups:

• Use session identity when selecting a group’s sender mark.
• Capture the gallery at this head for visual verification.

[cmux-bot]

cmux-bot approves 661da86: an independent review passed this exact head (router-review receipt by teamleaderleo). A push needs a new review; a later BLOCK at this head dismisses this approval.