acpmux: review boxes get no inbox notice and no other agent's steer by teamleaderleo · Pull Request #28369 · manaflow-ai/cmux

teamleaderleo · 2026-10-30T14:20:35Z

Summary

Review boxes started by lane-place new --review (harness codex-review-box) were getting the inbox notice from #27686 ("[inbox] New messages from other agents in N thread...") steered into their review turn. A coordinator found: "The reviewer then reads the thread instead of reviewing and ends with no SHIP/BLOCK verdict."

With this change a review box has no inbox, and no other agent's message steers into its turn:

review_box.rs: one predicate for "is a review box" (harness ends with -review-box, or a live review. tag). acpmux ls (cli/ls_table.rs) now uses it too, so the two can't drift apart.
hub/inbox.rs inbox_route: a review box takes no inbox item. A message to it goes the way it did before acpmux: every chat's inbox; another agent's plain send is never a turn of its own (A1) #27686.
hub/turns.rs: another agent's message to a review box's running turn never steers in, even though agent sends steer by default. It queues behind the review and runs as a turn of its own after the review ends. That turn's answer goes back to the sender's session/prompt, so no sender is left waiting. A steer-only send gets an error with reviewBox: true. The person's own steer and Send now still go in.
hub/inbox.rs inbox_review / notice_held: no inbox notice steers into a review box's turn. Messages its inbox took before it became one (a review. tag set later) wait for the turn's end, as an idle chat's do. notice_held is also checked as the steer is written (steer_end.rs).
hub/inbox_digest.rs: session inbox --digest refuses a review box (inbox_digest.review_box), so each agent message stays its own prompt.
A queued message the person sends with Send now still goes into the turn. That's the person's explicit action, like the person's own steer. The rule is that no other agent's message steers in.
Queued entries start with steer: false, and only the person's Send now changes that. So the queued path never steers into the running turn.
The module docs at the top of inbox.rs say review boxes are left out, and why.

Follows #27686.

Testing

Regression tests in tests/hub_integration/agent_inbox.rs, each committed before its fix:

a_review_box_takes_no_inbox_message_and_no_steer_during_its_review: an agent message to a review box during its turn makes no inbox item, no notice and no steer. It runs as its own turn and the sender gets end_turn. A steer-only send and a digest are refused. The same message to an ordinary chat still goes to that chat's inbox.
a_review_box_runs_its_first_prompt_from_another_agent_as_its_task: the task lane-place sends a *-review-box session as another agent runs as that prompt, with no notice in its place.
an_inbox_message_from_before_the_review_tag_never_steers_into_the_review.
Unit test a_review_box_harness_or_a_review_tag_marks_one.

Command: cmux-ci test -- cargo test -p acpmux --test hub_integration agent_inbox (plus --lib review_box, --lib inbox, --lib ls_table, inbox_digest and cargo clippy -p acpmux --all-targets -- -D warnings).

Observed / Not run

Observed: rustfmt is clean on the changed files. At 1db550a, clippy -D warnings passed and all 7 agent_inbox tests passed, the new one included (fast-checks runs 41071111850 and 41071584201). The results at the final head are posted as a PR comment.
Not run: the red run at the test-only commits. cmux-ci on a lane VM runs only at the PR's head, so the base failure is from reading the code, not from a run.

Changelog

Fixed: acpmux review boxes no longer get other agents' messages steered into their review.

Agent session: agent session

🤖 Generated with Claude Code

…ring its review Red on base: another agent's message to a review box's running turn goes to its inbox and its notice steers into the review. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

A review box (harness *-review-box, or a review. tag) takes another agent's message the pre-inbox way: queued behind the running review, a turn of its own after it, which answers the sender. No inbox item, no notice steered into the review, no agent steer and no digest steer. A steer-only send is refused with reviewBox. One predicate (review_box.rs) is shared with acpmux ls. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

blacksmith-sh · 2026-10-30T14:23:40Z

Found 6 test failures on Blacksmith runners:

Failures

Test	View Logs
`main.DeveloperIdAppTests.test_download_prints_the_active_profile_only/` `test_download_prints_the_active_profile_only`	View Logs
`stubs: /System/Library/Frameworks/Security.framework/Versions/A/Security has no stub/` `Security has no stub`	View Logs
`stubs: CoreFoundation.tbd: install-name /System/Library/Frameworks/` `CoreFoundation.framework/Versions/A/CoreFoundation, missing 1: ['_CFRetain']/` `CoreFoundation, missing 1: ['_CFRetain']`	View Logs
`stubs: CoreFoundation.tbd: install-name /System/Library/Frameworks/` `CoreFoundation.framework/Versions/A/CoreFoundation, missing 1: ['_log2']/` `CoreFoundation, missing 1: ['_log2']`	View Logs
`test_every_listed_hook_is_one_glaeda_released_with_the_check (main.SigningHooksInSt` `epWithGlaedaTests.test_every_listed_hook_is_one_glaeda_released_with_the_check)/` `test_every_listed_hook_is_one_glaeda_released_with_the_check (main.SigningHooksInSt` `epWithGlaedaTests.test_every_listed_hook_is_one_glaeda_released_with_the_check)`	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.}

… as its task lane-place sends a review box its task as another agent (acpmux new --prompt-file); on base it landed in the inbox and the box got a notice in its place. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

teamleaderleo · 2026-10-30T14:28:21Z

Owned by an agent session · session

…ers into the review Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

Review finding at 1db550a: notice delivery had no review-box check, so messages a chat's inbox took before a review. tag was set could still be noticed into its review turn. inbox_review and notice_held (also checked as the steer is written) now hold notices while a review box's turn runs; they go in after it ends, as an idle chat's do. A queued message the person sends with Send now still goes in: documented as the person's choice. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

cursor · 2026-10-30T14:40:45Z

Bugbot needs on-demand usage enabled

Bugbot uses usage-based billing for this team and requires on-demand usage to be enabled.

A team admin can enable on-demand usage in the Cursor dashboard.

Review finding at 5bc9922: session inbox --digest bundled a review box's queued agent prompts into one digest turn. It now refuses a review box (inbox_digest.review_box), so each waits as its own prompt behind the review. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

github-actions · 2026-10-30T14:48:15Z

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

Review 848bee8: 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=pr28369 -f ref=lane/review-box-inbox -f platform=macos -f backend_mode=local

Review finding at 74a882f: an agent message sent with interrupt (and a family message delivered with interrupt) still cancelled the running review turn. Both now queue behind it; the test checks the review ends end_turn. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

cursor · 2026-10-30T14:50:47Z

Bugbot needs on-demand usage enabled

Bugbot uses usage-based billing for this team and requires on-demand usage to be enabled.

A team admin can enable on-demand usage in the Cursor dashboard.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

Review finding at efc662a: take_joined coalesced queued agent prompts, so a review box's messages ran as one digest turn after the review. They now run each as its own prompt, as the inbox_digest refusal says; the test asserts three turns again. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

…e review Review finding at 2afd067: the test's FYI only armed the 10-minute flush, so base sent no notice and the test passed there. Now the message comes while the person's prompt waits behind a turn, so its notice is held for the next turn's start (base steers it into the review); the tag lands before that turn. Events, not sleeps: the review's answer, then the sender's. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

cursor · 2026-10-30T15:12:43Z

Bugbot needs on-demand usage enabled

Bugbot uses usage-based billing for this team and requires on-demand usage to be enabled.

A team admin can enable on-demand usage in the Cursor dashboard.

…rrupt Review findings at e57bc34: - an agent interrupt that saw no running turn kept its interrupt and could stop a review that started meanwhile: the guard no longer needs a turn; - a review. tag set while an agent steer or a family interrupt was on its way was not seen: the steer is checked again under the input handoff (a held one falls back to a turn of its own), and the family interrupt checks again before it cancels; - the pre-tag test could drop the answers it waits for: it reads all three together and waits on the FIFOs, not on an update. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

…ratchet) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-d9e7e9f7fb2b87a57d5f

…ing_slice 97 > 94): rewrite the new hits without panics (#28386) * acpmux: inbox, send and outbox status JSON writes without indexing Part of the acpmux panic ratchet base red that came back after #28134 (indexing_slicing 703 > 693, string_slice 97 > 94). These writes go through JsonSet::set instead of IndexMut. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-d9e7e9f7fb2b87a57d5f * acpmux: ls globs, dotted fields, inbox threads and match counts without indexing (10 hits from #28012, #27686, #27791, #27970) - cli/select.rs (911f25f, #28012): field() cuts the dotted path with str::get; glob() compares and tails with slice get. Every index was in bounds, so the output is unchanged. - hub/inbox.rs (cbe8e60, #27686): first_pr reads the number after the marker with str::get. inbox_unsend matches on (index, noticed, notice is none), with the same three outcomes in the same order. - hub/mod.rs (fe3152a, #27791): local_match_counts bumps its slot with get_mut. - hub/queue.rs (46ed428, #27970): the transfer id is read with get. Tests add multibyte names, keys and PR markers. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-d9e7e9f7fb2b87a57d5f * acpmux: ACP session/list paging without indexing (5 indexing_slicing hits from #28161) #28161 (4543835) landed while this PR was in review: - hub/session_pages.rs: byte_page's sort key reads each key with get. Every index comes from enumerate, so the order is unchanged. - server/list_reply.rs: sessions and updatedAt are read with get (a missing value is Null, as before), and nextCursor is written via JsonSet::set. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-d9e7e9f7fb2b87a57d5f * panic ratchet: lower acpmux to the measured counts (indexing_slicing 731 -> 686, string_slice 97 -> 94, unwrap_used 144 -> 143) #28169 raised the acpmux baseline to the counts base had then, so the hits fixed here could come back unnoticed. Measured at fdb6218 (fast-checks run 41080915558). PR CI now runs the ratchet too (#28169), so a PR that adds a hit is caught before it lands. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-d9e7e9f7fb2b87a57d5f * acpmux: read ls row tags with get (the hit #28369 added to the panic ratchet) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-d9e7e9f7fb2b87a57d5f --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

github-actions · 2026-10-30T16:59:06Z

Shipped in:

cmux NEXT nightly 1.0.0-nightly.4108762076801 (8e3aaa2401)
cmux NEXT nightly 1.0.0-nightly.4108980870001 (d9f2f7af55)
cmux NEXT nightly 1.0.0-nightly.4109063053901 (bf70709747)
cmux NEXT nightly 1.0.0-nightly.4109296117601 (72882288da)
cmux NEXT nightly 1.0.0-nightly.4109410702301 (aead767cca)

… changes (#28699) cmux-tui-spec.yml's clippy job judges a PR's Rust against the rust-panic-baseline.json rows at its own merge. A row lowered on the base after that left the green standing, so #28182 (6 acpmux indexing hits) and #28369 (1) landed over rows lowered since their checks ran. The baseline is already a check path, so a merger dispatches the fast gates for such a PR. On a dispatch, prepare now writes ratchet=true when the tip lowered a deny-lint row of a crate the PR changes (any row, for a workspace-wide Rust input) since the merge base. The new panic-ratchet job then lints those packages at the tip and on the merge (cmux-tui-clippy-changed.py --compare-base) and the ratchet (--base-clippy-json) fails only on counts over the tip's row that the tip does not already reach, so a base ratchet red never reads as the PR's. post fails the status unless that job passed or was skipped. Agent-Session: https://cmux.com/deeplink/agent?id=aid-7e68dca28911d1051663 Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

teamleaderleo and others added 2 commits October 30, 2026 14:20

teamleaderleo added the risk:normal Risk tier (docs/ci/risk-tiers.md) label Oct 30, 2026

teamleaderleo temporarily deployed to sentry-dsyms October 30, 2026 14:26 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to artifacts October 30, 2026 14:26 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to artifacts October 30, 2026 14:27 — with GitHub Actions Inactive

teamleaderleo and others added 2 commits October 30, 2026 14:40

acpmux: test that an inbox message from before a review tag never ste…

e84d32a

…ers into the review Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

teamleaderleo temporarily deployed to ci-routing-agent October 30, 2026 14:47 — with GitHub Actions Inactive

teamleaderleo and others added 2 commits October 30, 2026 14:58

acpmux: the review-box test lets queued agent messages join one turn

3af861c

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Agent-Session: https://cmux.com/deeplink/agent?id=aid-3db671e49c5e0014c84a

teamleaderleo temporarily deployed to ci-routing-agent October 30, 2026 15:08 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to ci-routing-agent October 30, 2026 15:29 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to artifacts October 30, 2026 15:37 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to sentry-dsyms October 30, 2026 15:37 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to artifacts October 30, 2026 15:38 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to artifacts October 30, 2026 15:46 — with GitHub Actions Inactive

teamleaderleo temporarily deployed to artifacts October 30, 2026 15:48 — with GitHub Actions Inactive

teamleaderleo merged commit 364c30d into feat-cmux-next Oct 30, 2026
187 of 190 checks passed

teamleaderleo deleted the lane/review-box-inbox branch October 30, 2026 15:59

teamleaderleo mentioned this pull request Oct 30, 2026

acpmux: a chat's first task from another agent runs as its prompt, not an inbox notice #28413

Merged

teamleaderleo mentioned this pull request Oct 30, 2026

Base red: acpmux panic ratchet again (indexing_slicing 703 > 693, string_slice 97 > 94): rewrite the new hits without panics #28386

Merged

Navigation Menu

Repository navigation

acpmux: review boxes get no inbox notice and no other agent's steer - #28369

acpmux: review boxes get no inbox notice and no other agent's steer#28369
teamleaderleo merged 11 commits into
feat-cmux-nextfrom
lane/review-box-inbox

teamleaderleo commented Oct 30, 2026 •

edited

Loading

Uh oh!

blacksmith-sh Bot commented Oct 30, 2026 •

edited

Loading

Uh oh!

teamleaderleo commented Oct 30, 2026

Uh oh!

cursor Bot commented Oct 30, 2026

Uh oh!

github-actions Bot commented Oct 30, 2026 •

edited

Loading

Uh oh!

cursor Bot commented Oct 30, 2026

Uh oh!

cursor Bot commented Oct 30, 2026

Uh oh!

Uh oh!

github-actions Bot commented Oct 30, 2026 •

edited

Loading

Uh oh!

This branch was previously deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

teamleaderleo commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Testing

Observed / Not run

Changelog

Uh oh!

blacksmith-sh Bot commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Failures

Uh oh!

teamleaderleo commented Oct 30, 2026

Uh oh!

cursor Bot commented Oct 30, 2026

Bugbot needs on-demand usage enabled

Uh oh!

github-actions Bot commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cursor Bot commented Oct 30, 2026

Bugbot needs on-demand usage enabled

Uh oh!

cursor Bot commented Oct 30, 2026

Bugbot needs on-demand usage enabled

Uh oh!

Uh oh!

github-actions Bot commented Oct 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This branch was previously deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

teamleaderleo commented Oct 30, 2026 •

edited

Loading

blacksmith-sh Bot commented Oct 30, 2026 •

edited

Loading

github-actions Bot commented Oct 30, 2026 •

edited

Loading

github-actions Bot commented Oct 30, 2026 •

edited

Loading