Windows Companion: "Agent thinking" animation closes the chat/session selectbox before a different chat can be chosen

[danilovmy]

When the animated Agent thinking indicator is visible, it becomes difficult or impossible to switch to another chat from the selectbox because the animation keeps triggering UI updates/repaints and the selectbox closes before the user can complete the selection.
In practice this makes switching chats unreliable or impossible while the animation is active.

Why this matters

Switching chats while one agent is still thinking is a normal workflow. The animated status should not steal focus or destabilize the session/chat picker.

Repro

1. Open Windows Companion.
2. Start or wait for a chat where the UI shows the animated Agent thinking label.
3. Open the chat/session selectbox.
4. Try to choose another chat while the animation is still running. In practice this makes switching chats unreliable or impossible while the animation is active.

Expected

• The selectbox should stay open long enough for the user to pick another chat.
• The Agent thinking animation can continue visually, but it must not reset/close the open dropdown.
• Ongoing animated status updates should not interfere with normal pointer/keyboard selection in the chat picker.

Actual

• The screen keeps repainting while Agent thinking animates.
• The selectbox closes before the selection can be completed.

Suggested fix

• Ensure the Agent thinking animation does not remount or reset the chat/session picker.
• Keep the picker open state stable across unrelated animation/status updates.

[clawsweeper]

Codex review: keeping this open for maintainer follow-up; there is still a little grit to resolve. Reviewed July 11, 2026, 4:17 PM ET / 20:17 UTC.

Summary
Current main still replaces the native chat-session ComboBox during snapshot-driven renders, matching the reported flyout closure while thinking updates arrive. The report is focused, source-reproducible, and not covered by the adjacent session-persistence or large-list items.

Reproducibility: yes. at source level: provider snapshots request rerenders, and each render replaces the session ComboBox identified in the screenshot. A current-head native WinUI recording remains the appropriate runtime confirmation.

Next step
The defect is narrow, current-main source exposes the replacement path, and the likely implementation, test, and proof surfaces are clear.

Review details

Best possible solution:

Keep the grouped session picker as a stable reconciled control, update its items and selected value without replacing it, and prove its flyout remains open through repeated thinking-status renders.

Do we have a high-confidence way to reproduce the issue?

Yes at source level: provider snapshots request rerenders, and each render replaces the session ComboBox identified in the screenshot. A current-head native WinUI recording remains the appropriate runtime confirmation.

Is this the best way to solve the issue?

Yes directionally; stabilizing the picker instance is narrower and safer than suppressing legitimate thinking or timeline renders, but the fix should include a regression seam and live UI proof.

AGENTS.md: found and applied where relevant.

Remaining risk / open question:

• The exact live update frequency still needs current-head WinUI confirmation, although the source replacement mechanism and reporter screenshot strongly support the report.

Codex review notes: model internal, reasoning high; reviewed against 622c0e27c6b0.

Label changes

Label changes:

• add P1: A normal active-chat workflow becomes unreliable or unusable while an agent is thinking, and current source exposes the render-driven replacement path.
• add impact:ux-friction: The live status animation interferes with switching among existing chats, creating avoidable user-facing friction.
• add issue-rating: 🦞 diamond lobster: Current issue advisory state selects this label.
• add clawsweeper:source-repro: Current issue advisory state selects this label.
• add clawsweeper:queueable-fix: Current issue advisory state selects this label.
• add clawsweeper:fix-shape-clear: Current issue advisory state selects this label.

Label justifications:

• P1: A normal active-chat workflow becomes unreliable or unusable while an agent is thinking, and current source exposes the render-driven replacement path.
• impact:ux-friction: The live status animation interferes with switching among existing chats, creating avoidable user-facing friction.

Evidence reviewed

Acceptance criteria:

• ./build.ps1
• dotnet test ./tests/OpenClaw.Shared.Tests/OpenClaw.Shared.Tests.csproj --no-restore
• dotnet test ./tests/OpenClaw.Tray.Tests/OpenClaw.Tray.Tests.csproj --no-restore
• Launch ./run-app-local.ps1 -Isolated and record opening the session picker, keeping it open through repeated thinking updates, and selecting another session

What I checked:

• Snapshot-driven rerender: The chat root stores every provider snapshot change, including status-only changes, which requests a new render. (src/OpenClaw.Tray.WinUI/Chat/OpenClawChatRoot.cs:172, 622c0e27c6b0)
• Picker recreation: The composer creates a new native ComboBox inside a Border setter and assigns it as the child on each render instead of updating a stable picker instance. (src/OpenClaw.Tray.WinUI/Chat/OpenClawComposer.cs:252, 622c0e27c6b0)
• Renderer behavior: FunctionalUI reuses controls by path but reruns setters on every render, so the stable Border receives the newly constructed ComboBox child. (src/OpenClawTray.FunctionalUI/FunctionalUI.cs:1021, 622c0e27c6b0)
• Reporter evidence: The attached screenshot identifies the native session dropdown directly beneath the active assistant-thinking indicator, matching the inspected control.
• Feature history: Recent merged work on native session selection and chat rendering points to shared ownership across the chat root, composer, and FunctionalUI renderer. (src/OpenClaw.Tray.WinUI/Chat/OpenClawChatRoot.cs:148, c1bd9ae58b9e)

Likely related people:

• TurboTheTurtle: Authored the latest merged native session-selection persistence change and recent chat virtualization work in the central chat path. (role: recent chat contributor; confidence: high; commits: c1bd9ae58b9e, fcc74778a35c; files: src/OpenClaw.Tray.WinUI/Chat/OpenClawChatRoot.cs, src/OpenClaw.Tray.WinUI/Chat/OpenClawComposer.cs)
• calebeden: Authored recent FunctionalUI render-identity and chat-state changes adjacent to control reconciliation and rerender behavior. (role: recent renderer contributor; confidence: high; commits: f89a88a6baf4, 02f2421a5e75, c9b5476a873f; files: src/OpenClawTray.FunctionalUI/FunctionalUI.cs, src/OpenClaw.Tray.WinUI/Chat/OpenClawChatRoot.cs, tests/OpenClaw.Tray.Tests/ChatTimelineRenderIdentityContractTests.cs)
• Scott Hanselman: Available blame attributes the grouped native session picker and baseline FunctionalUI control-reuse path to the v0.6.12 baseline commit. (role: baseline implementation contributor; confidence: medium; commits: 4166e0fd63f8; files: src/OpenClaw.Tray.WinUI/Chat/OpenClawComposer.cs, src/OpenClawTray.FunctionalUI/FunctionalUI.cs)

How this review workflow works

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

[clawsweeper]

🦞⚠️
Automatic implementation stopped before completion.
Reason: The automatic implementation worker stopped before all deterministic gates completed. Open the workflow run for the exact blocker.
Worker: https://github.com/openclaw/clawsweeper/actions/runs/29173264273

[shanselman]

Confirmed fixed on current main by the already-merged #991 and #984. Current FunctionalUI preserves the native ContentFlyout by render path while OpenClawChatRoot continues legitimate snapshot/status rerenders. Added real WinUI regression coverage in #1035: repeated Agent thinking-style renders preserve button/flyout/content identity and IsOpen; rows and selection update; mark-and-sweep closes a removed picker. Mutation proof replacing path caching with a new Flyout per render makes the test fail at native identity, and restoring production returns it green. Current-head visible proof showed Agent thinking 3 and the open native Session popup simultaneously. Required build, Shared (2,925 passed/31 skipped), Tray (1,823 passed), FunctionalUI (19 passed), and focused WinUI (1 passed) validation all passed.