fix(orchestration): native-chat agents get their mail notice right away, the same way for every agent by brennanb2025 · Pull Request #28540 · stablyai/orca

brennanb2025 · 2026-10-14T03:50:31Z

	Files	Added	Deleted	Net
Test	8	$\color{#1a7f37}{\Huge{\mathbf{+}}}$195	$\color{#cf222e}{\Huge{\mathbf{−}}}$269	$\color{#cf222e}{\Huge{\mathbf{−}}}$74
Prod	8	$\color{#1a7f37}{\Huge{\mathbf{+}}}$157	$\color{#cf222e}{\Huge{\mathbf{−}}}$80	$\color{#1a7f37}{\Huge{\mathbf{+}}}$77

ELI5

When a worker agent reports back to a coordinator running in a native chat, Orca posts a short "You have 1 orchestration message, run check" notice into the coordinator's chat. If the coordinator was busy, that notice waited as a queued card until the coordinator's whole turn ended. A coordinator that spends its turn in a long command (for example, a loop polling its own inbox) never saw its workers' reports until it stopped, and the cards piled up. Now the notice is simply sent to the agent right away, the same way for every agent, just as a terminal agent gets its notice typed in.

What Changed

Before: every notice to a busy native chat became a queued card ("From … Steer / Delete") and was only sent after the turn ended. Several reports during one turn meant several cards.

After: every native-chat agent gets the notice sent right away, exactly as if you had typed it and pressed send. No card is created, for any agent. What the agent then does with a message that arrives mid-turn is the same thing it does with your own mid-turn message:

Claude, Codex and Pi take it into the running turn when their current step (usually a command) finishes.
Grok, OpenCode and the other ACP agents stop the running prompt and answer the message as the next prompt. This is what these agents already do when you send while they work; the coordinator picks its work back up in its reply.
Terminal agents: no change. They already work this way, since the notice is typed into the agent's own input.

One hold: while the chat is waiting on you to answer an approval or a question, the notice is not sent. Sending then would land in the turn you are answering, and on ACP agents would withdraw the open prompt you were about to answer. The notice goes out the moment you answer, or when the turn ends. Held mail never shows as a card or a bubble; it simply arrives after you answer.

Mechanism:

The notice used to be sent with delivery: 'queue' (queue as a card if busy). It is now always a plain send (delivery: 'now') for every agent. Delivery is part of what the host checks when a send is retried under the same operation ID, and it no longer depends on the agent or the moment, so a retry can never disagree with the first attempt.
The hold lives where the send is accepted. When the chat session admits a mail notice, it checks, in the same step, whether an approval or question is waiting on the person. If one is, it refuses the notice before anything is recorded: no submission, no receipt, no card. The mail stays unread and parked. A person's own sends are admitted at the same place and are never held.
Parked mail is retried when the chat goes back to working (the edge that follows your answer) or goes idle. The park is registered before the sender reads anything, and after a refusal the sender re-checks once, so an answer that lands mid-attempt is never missed.
An earlier version checked for the prompt in the mail sender, before the send. Review reproduced two timing bugs in that: a prompt opening just after the check could be withdrawn on ACP agents, and an answer landing just before the park could leave mail waiting until the turn ended. Moving the check into the admission step removes both.

Why

The queue was chosen so agent mail would behave like a person's queued message. For a coordinator, though, "wait for the turn to end" can mean waiting until the coordinator gives up. Sending right away matches what terminal coordinators already get, through the one send path native chat already has, and behaves identically for every agent so users see one behaviour. Holding while a prompt waits on the person follows the common pattern: a notice for an agent is held while one of its prompts waits on the person, and goes out after the answer.

Alternatives considered:

Send right away only to agents that take a message mid-turn, and keep cards for ACP agents (this PR's first version). Rejected: two behaviours for one feature confuse users.
Letting the sending agent choose to steer per message (feat(orchestration): let an agent steer its message into a busy chat instead of queueing it #26110). This keeps queueing as the default, so the common case stays slow. If this lands, feat(orchestration): let an agent steer its message into a busy chat instead of queueing it #26110 can be closed.

Differences from the common pattern:

Intended: on agents that stop the running prompt for a new message, a mail notice does too, as your own message does. Keeping that consistent with a person's send is the point of the change; a notice never arrives while you are answering a prompt, so no open prompt is withdrawn.
Intended: the hold is decided by the chat session itself at the moment it accepts the send, and nothing is stored: held mail stays unread, and the next working or idle edge, plus one immediate re-check, sends it.
Intended: a tiny window remains between the chat accepting the notice and the agent receiving it, in which a new prompt could open. A person's own message has exactly the same window.

This is PR 1 of 2. PR 2 replaces the notice with the actual mail content for native chats, one chat message per mail message, so the coordinator no longer needs to run check.

Linked Issue

None (maintainer change).

Visual Proof

Rendered-app QA on this change (b4cd51c5714), with real signed-in agents (Codex and Grok) in a hidden test copy on a separate Mac:

1. Busy Codex coordinator: the worker's notice arrives as a sent message (this PR)

Same flow on main: the notice waits as a queued card

2. Codex waiting on an approval: mail is held, then sent right after Allow

3. Grok (an ACP agent): busy chat takes the notice next; mail held around a question and released after the answer

Testing

I manually tested these changes locally
Automated tests added/updated, or explained why not below
End-to-end test on the coordinator-mail test rig: a worker report that arrives while the coordinator is mid-turn goes into that turn as a Codex turn/steer naming the running turn. No card is created, the message keeps who it's from, and nothing is resent at the idle edge.
Real host plus ACP test: with a question already open, a mail notice is refused with nothing recorded, the question stays answerable, and exactly one plain send happens after the answer.
Delivery test: the park is registered before the session read, and an answer landing during that read still leads to one send, with no duplicate and no lost unread message.
Host test: the notice is a plain send for every agent. Status wiring test: a chat going back to working retries its parked mail.
Removed structured-chat-coordinator-mail-queue.test.ts: all four of its tests asserted the queue-until-turn-end behaviour this PR replaces. The 20-card queue limit it also covered is still tested at the host level.
The 52 targeted tests pass under Node (pnpm test runs them under Bun, where node:sqlite is unavailable, on main too). CI runs lint, typecheck and the full suite.

Notes

SSH and remote: the hold is decided on the host that runs the chat, inside that chat's own session. Nothing new crosses the wire and no new saved field. An old paired server keeps its current behaviour until it updates.
The final sync with main let the commit hook reformat three files from main (structured-agent-session-send-runtime.ts, structured-agent-session-send-settlement.ts, AiVaultScanProgressStrip.test.tsx); those changes are whitespace only.

brennanb2025 · 2026-10-14T04:23:21Z

A busy native-chat coordinator previously waited until its entire turn ended to see worker-mail notices, because they became queued cards. This PR sends the notice through the ordinary immediate-send path for every agent. Agents that restart for a new message also restart for mail, just as they do for a person's message; terminal delivery is unchanged. Mail remains unread while the host sees a pending approval or question, without creating a queued card or an early chat message.

The fresh review reproduced two races in that hold: a prompt opening after an earlier snapshot could be cancelled by mail, and an answer arriving before retry registration could leave mail waiting for an unrelated event. Commit b4cd51c5714ec9c4590d192968a228c058a4bc9b checks the host's live prompt state inside existing send admission, before accepting a chat message, and registers the existing retry before asynchronous reads. A live release check covers answers arriving during a held attempt. The existing working and idle status changes remain the settlement wake. The earlier correction that drains later reports after successful acceptance is preserved. No saved hold flag, timer, provider-specific sender, or wire change was added.

The common pattern uses current-owner admission and rechecks remaining human interactions after settlement. Intended differences: all providers use ordinary immediate send, including interruption where that is how a person's send behaves; held mail stays invisible and unread. A prompt opening after host admission retains the same provider-handoff timing window as a person's ordinary send. This review does not claim stronger provider-level atomicity.

Fresh correctness and readiness source reviews found no remaining proven defect. The final Sol xhigh decider independently traced the final production path and withdrew the earlier retry-loop and map-fanout concerns: no correctness defect or material resource regression is proven. Repeated prompt activity and large concurrent mailbox/history workloads remain deferred, unmeasured scale concerns.

Verified on the committed correction: four targeted Node suites, 52 tests passed, and the mandatory commit hook passed after consolidating duplicate missing-session/fence handling and redundant park ownership checking. The ACP host contract proves that a pending question holds mail without accepting a message, writing a send receipt, or cancelling the prompt. A delivery-lane contract proves one release after an answer arrives during the held send; a separate production-host fixture traces a real answer and one notice. These automated fixtures do not establish a complete real ACP provider exchange or rendered behavior.

CI: PR Checks run 38816106437 has static analysis and typecheck passing. The Windows package job failed in the existing installer process probe at config/scripts/nsis-process-check.test.mjs:102 after a timed-out SIGTERM; the package and Node test jobs were still queued when the fail-fast watch exited, so the aggregate is not green and the failure needs infrastructure or test-owner follow-up.

Real-agent Electron QA is blocked. The M4Air and Windows attempts ended before launching Orca because no authorized signed-in agent was available in the isolated test homes. No current-head screenshots or real-provider rendered flows were produced. Busy native delivery, approval/question hold and post-answer delivery, and ACP behavior still need authenticated remote QA; older fixture screenshots below cover an earlier revision only. The PR remains a draft and is not ready to merge.

Historical controlled-provider screenshots — old head 4f1d256, not current-head QA

These images were captured with an isolated Codex fixture on M4Air during the earlier design review. They show the queued baseline and immediate notices, but do not validate the final human-prompt hold or authenticated providers.

Nothing was merged or synchronized with main during this correction. The separate mail-content PR #28567 remains outside this review and unready.

QA on `b4cd51c`

Real signed-in Codex (GPT-6.1-Sol) and Grok ACP (Grok 4.7), driven by grok -p on M4Air on October 16, 2026, using auth-only copies in dedicated QA homes and hidden CDP renderers. This completes the authenticated rendered QA that the earlier blocked status above lacked.

Flow 1 — exercised: during a real 180-second Codex command, the PR sent worker mail immediately as a plain notice with no queued card; origin/main at checkout (f73ab77b46c) showed the queued From/Steer/Delete card.
Flow 2 — exercised: the same Codex chat held new mail invisibly while a real command approval stayed answerable; after Allow, the notice appeared in the next screenshot 1.8 seconds later.
Flow 3 — exercised: Grok ACP interrupted its running prompt for a plain mail notice; its real question kept mail invisible until Green was selected, then displayed the notice in the next screenshot 2.2 seconds later.

The Grok question proof used a fresh Run on the same chat, excluding an outstanding unacknowledged delivery from the earlier busy test. Mail came from real test-app shell workers through their own Orca CLI; no provider fixtures or injected rows were used.

Busy native chat: PR before / after; main before / after

Native approval: before mail / held mail / after answer

Grok ACP busy: before / after

Grok ACP question: before mail / held mail / after answer

Both QA source worktrees, throwaway homes/profiles, test apps, and daemons were removed; their ports have no listeners. No PR source was edited, pushed, or merged. No lint or typecheck command was run.

Real-file hashes are reported without attributing concurrent shared-machine changes: real Codex config.toml and .claude.json changed; neither contained a QA path when inspected, and all other sampled files matched.

Real configuration SHA-256 before / after

Real file under machine home	Before	After
`.codex/auth.json`	`1283b66d6abcb91796a3fb86321f21713fb91b67de053b8eca96153f8ca42322`	`1283b66d6abcb91796a3fb86321f21713fb91b67de053b8eca96153f8ca42322`
`.codex/config.toml`	`e900c9107ea5074eb8425d45248a9a79ec9e518f3b53473dfa478654c919ba53`	`1cb3428af386e74d64aed356d18f81cabe35b4bcebcf1c0aca58675de76ed6c1`
`.codex/hooks.json`	`edb2d1cab54568e86a32aa813939d8a8892f1184a0d13bddbd08e9059a723eee`	`edb2d1cab54568e86a32aa813939d8a8892f1184a0d13bddbd08e9059a723eee`
`.grok/auth.json`	`cbc8c26d87d1bccb024c718c7988e9094ec1d22d1708dd1584b62cdc6f0e905a`	`cbc8c26d87d1bccb024c718c7988e9094ec1d22d1708dd1584b62cdc6f0e905a`
`.grok/config.toml`	`d0b1d13307504f7d888ba40924942b75fb90e1572b9314e7a103b2d729de74b1`	`d0b1d13307504f7d888ba40924942b75fb90e1572b9314e7a103b2d729de74b1`
`.claude.json`	`ec676927413b47f6af3a8de4d507e6928935bbdb1c180ec0a0ef71c8a6a8372e`	`f3baf8b37daae16aaafdd26e0b0580caea99b24dd6cd6c2af0c57e8da8e4bbb1`
`.claude/settings.json`	`7847561ee31422545b5f29fc9154aeb4b73733c46e0457b719d1b633920a45c0`	`7847561ee31422545b5f29fc9154aeb4b73733c46e0457b719d1b633920a45c0`
`.claude/.credentials.json`	`absent`	`absent`
`.config/opencode/cli.json`	`d0deb7922566a96051665a87221e568ced788f3059888b11f27cace831bde3e0`	`d0deb7922566a96051665a87221e568ced788f3059888b11f27cace831bde3e0`
`.config/opencode/service.json`	`5ff5f8d3b6ac674c426b68d2a394755f9907d2cb0a7c550986caf73a1066ead9`	`5ff5f8d3b6ac674c426b68d2a394755f9907d2cb0a7c550986caf73a1066ead9`
`.local/share/opencode/auth.json`	`absent`	`absent`

This is rendered QA evidence only; it does not change the CI merge gate or mark the PR ready.

Final sync (`4495dbd`)

Main merged at the end (current main, after #29809 fixed main's lint break from #28729). The merges add only main's changes plus whitespace-only reformatting of three main files by the commit hook; no change to this PR's logic since the reviewed head b4cd51c5714. Review loop: done, no open findings. QA: passed (above).

…ing turn A native-chat coordinator's mail notice always queued as a card until the whole turn ended, so a coordinator busy in a long command never saw its workers' reports. Agents that take a message mid-turn (Claude, Codex, Pi) now get the notice at once; agents that would cancel their turn for it (ACP) still queue it.

coderabbitai · 2026-10-15T20:13:07Z

📝 Walkthrough

Mailbox delivery now processes successive unread batches after successful handoffs and pauses while a session awaits a human response. Structured-session pointers use immediate delivery. Working-status events retry parked mail before mailbox redrive. Coordinator tests cover steering pointers into active turns, and the fake Codex fixture records steering requests. The mail-queue integration test suite was removed.

Priority: ⬇️ Low

Merge Risk: ⚪ Minimal · up to 4495d

Structured mail is held while a human response is pending, but automated tests do not cover the answer-triggered coordinator retry end to end. No production failure is established, and the reported Windows package timeout is isolated from the changed code; code-level merge risk is minimal.

Pre-merge checks: 4 · 1

Failed checks: 1 warning

Check name	Status	Explanation	Resolution
Docstring Coverage		Docstring coverage is 54.55% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 11 functions across 15 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

Passed checks: 4 passed

Check name	Status	Explanation
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.
Title check		The title clearly identifies the primary change: native-chat orchestration mail notices are delivered immediately and consistently across agents.
Description check		The description is detailed and covers the user impact, implementation, rationale, alternatives, testing, visual proof, and compatibility notes. It omits some template sections, such as the checklist …

Full details: Description check

Explanation

The description is detailed and covers the user impact, implementation, rationale, alternatives, testing, visual proof, and compatibility notes. It omits some template sections, such as the checklist and agent skill boundary, but the required change information is substantially complete.

Fix all pre-merge checks with AI

✨ Finishing Touches: 💡 1

📝 Generate docstrings: 💡

Commit to this branch
Create a new PR

Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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

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

A mail notice is a plain send for every agent, as a person's own mid-turn message is. While a pending approval or question waits on the person it is held, and the turn's working edge after the answer sends it.

coderabbitai

Actionable comments posted: 1

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI
Review profile: CHILL
Plan: Advanced
Run ID: a968857f-25f4-4080-992a-0c51b19f7b9d

📥 Commits

Reviewing files that changed from the base of the PR and between 0bcfbbf and e1e584e.

📒 Files selected for processing: 7

src/main/runtime/orca-runtime-structured-status-sink-wiring.test.ts
src/main/runtime/orchestration/agent-facing-parity.test.ts
src/main/runtime/orchestration/structured-mailbox-pointer-delivery.test.ts
src/main/runtime/orchestration/structured-mailbox-pointer-delivery.ts
src/main/runtime/orchestration/structured-mailbox-pointer-host.test.ts
src/main/runtime/orchestration/structured-mailbox-pointer-host.ts
src/main/runtime/orchestration/terminal-pane-mail-routing.ts

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 6 remain after this review.

# Conflicts: # src/main/runtime/structured-chat-coordinator-mail-queue.test.ts

…ation-mail-steer

coderabbitai

🧹 Nitpick comments: 1

src/main/runtime/structured-chat-coordinator-mail.test.ts: 1

459-497: 🎯 Functional Correctness | 🔵 Trivial | ⚡ Quick win

Cover the answer-triggered retry on the coordinator path.

acp-structured-mail-admission.test.ts covers a real ACP question and held mail, but it stops at { kind: 'held' }. structured-chat-dispatch-worker-mail.test.ts covers provider failure and retry on a later turn. The coordinator test covers active-turn steering only.

Add a focused test that answers the provider question after mail is held and asserts one coordinator retry. This can catch answer-to-retry wiring regressions that the existing tests miss.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI
Review profile: CHILL
Plan: Advanced
Run ID: ff2fcaef-3ae7-4bbb-9b53-9a1fc8144c26

📥 Commits

Reviewing files that changed from the base of the PR and between 6d94baf and 4495dbd.

📒 Files selected for processing: 1

src/main/runtime/orchestration/send-agent-turn.ts

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 7 remain after this review.

…28540's hold Each orchestration message still reaches a native chat as its own message, now through the merged plain send for every agent and its admission hold: a message held for a waiting prompt stops the pass so later mail cannot overtake it, and the mailbox is parked before the first read. Heartbeats are skipped in this lane only and stay unread. A mailbox with an unacknowledged check batch waits for its ack, as on main. Chat text is capped at 4,000 characters with the command that shows the whole message.

…worker A coordinator's follow-up to a chat-mode worker (Chat UI on) always reached it as a queued card that waits for the worker's current turn to end. `orca orchestration send --deliver now` stores the request on the message row, and the structured pointer lane sends the batch's nudge with delivery `now`, which joins the running turn the way the composer's Steer button does. The default stays `queue`. - New `messages.pointer_delivery` column (schema v44, defaults to `queue`). - Delivery is read off the batch's own rows, so a retry under a reused operation id always resends with the delivery it was minted with (delivery is part of the host's operation fingerprint). - Carried through group sends and both federation relay directions as an optional field written only for `now`, so plain mail is byte-identical for older peers. - Terminal (PTY) recipients are unchanged: they are still nudged at the next idle prompt, since typing into a busy agent TUI can land in a permission prompt. - CLI help, flag help, and the orchestration messaging guide document it. Ported onto upstream stablyai#28540, which already steers every mail notice into a busy chat's running turn (held only while an approval or question waits). The pointer lane keeps that upstream behavior for every row; the `--deliver` flag, the v44 column and the federation field stay so existing scripts and databases keep working, and the help text says a row's `queue` no longer holds a nudge. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

brennanb2025 added 2 commits October 13, 2026 21:57

fix(orchestration): drain mail that arrives while a notice is pending

4f1d256

brennanb2025 mentioned this pull request Oct 14, 2026

fix(orchestration): native chats get each orchestration message whole instead of a check pointer #28567

Open

brennanb2025 force-pushed the brennanb2025/orchestration-mail-steer branch from e9394b6 to abbb2ea Compare October 14, 2026 05:05

brennanb2025 mentioned this pull request Oct 14, 2026

fix(native-chat): give the chat option write its clock (main typecheck) #28570

Closed

1 of 2 tasks

brennanb2025 force-pushed the brennanb2025/orchestration-mail-steer branch from abbb2ea to 4f1d256 Compare October 14, 2026 05:23

Merge main into orchestration-mail-steer (picks up #28571 clock fix)

0bcfbbf

brennanb2025 marked this pull request as ready for review October 15, 2026 20:10

Send orchestration mail the same way to every chat agent

e1e584e

A mail notice is a plain send for every agent, as a person's own mid-turn message is. While a pending approval or question waits on the person it is held, and the turn's working edge after the answer sends it.

brennanb2025 changed the title ~~fix(orchestration): a busy native-chat coordinator gets its mail notice mid-turn instead of after the turn~~ fix(orchestration): native-chat agents get their mail notice right away, the same way for every agent Oct 15, 2026

brennanb2025 marked this pull request as draft October 15, 2026 20:24

coderabbitai Bot reviewed Oct 15, 2026

View reviewed changes

Comment thread src/main/runtime/orchestration/structured-mailbox-pointer-host.ts Resolved

brennanb2025 added 2 commits October 15, 2026 15:17

Hold orchestration mail at live prompt admission

b4cd51c

Merge commit '42dfd813fed' into brennanb2025/orchestration-mail-steer

6d94baf

# Conflicts: # src/main/runtime/structured-chat-coordinator-mail-queue.test.ts

brennanb2025 marked this pull request as ready for review October 16, 2026 18:34

Merge remote-tracking branch 'origin/main' into brennanb2025/orchestr…

4495dbd

…ation-mail-steer

brennanb2025 added the cherry-pick label Oct 16, 2026

coderabbitai Bot reviewed Oct 16, 2026

View reviewed changes

brennanb2025 merged commit 702ffc7 into main Oct 16, 2026
33 checks passed

brennanb2025 deleted the brennanb2025/orchestration-mail-steer branch October 16, 2026 21:45

brennanb2025 mentioned this pull request Oct 19, 2026

feat(orchestration): let an agent steer its message into a busy chat instead of queueing it #26110

Closed

9 tasks done

AmethystLiang mentioned this pull request Oct 20, 2026

[Bug]: Queued orchestration heartbeat notices fire after the run is drained, triggering empty-check loops #30994

Closed

This was referenced Oct 22, 2026

[Feature]: Improved orchestration #4376

Closed

[Feature] Ability to have “steer” be the default for cross-agent messages? #28960

Closed

Tauri-EPO mentioned this pull request Oct 26, 2026

[Other]: Orchestration loop roadmap — decisions, dependencies and acceptance #15190

Open

Navigation Menu

Repository navigation

fix(orchestration): native-chat agents get their mail notice right away, the same way for every agent - #28540

fix(orchestration): native-chat agents get their mail notice right away, the same way for every agent#28540
brennanb2025 merged 7 commits into
mainfrom
brennanb2025/orchestration-mail-steer

brennanb2025 commented Oct 14, 2026 •

edited

Loading

Uh oh!

brennanb2025 commented Oct 14, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Oct 15, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

brennanb2025 commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

ELI5

What Changed

Why

Linked Issue

Visual Proof

Testing

Notes

Uh oh!

brennanb2025 commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

QA on b4cd51c

Final sync (4495dbd)

Uh oh!

coderabbitai Bot commented Oct 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

brennanb2025 commented Oct 14, 2026 •

edited

Loading

brennanb2025 commented Oct 14, 2026 •

edited

Loading

QA on `b4cd51c`

Final sync (`4495dbd`)

coderabbitai Bot commented Oct 15, 2026 •

edited

Loading