Deflake CI E2E failures (#3311)

e2e-tests/edit_code.spec.ts

@@ -39,6 +39,10 @@ async function replaceEditorContent(page: Page, content: string) {
   await page.keyboard.type(content);
 }
 
+function normalizeLineEndings(value: string) {
+  return value.replace(/\r\n/g, "\n");
+}
+
 test("edit code", async ({ po }) => {
   await po.setUp({ autoApprove: true });
   const editedFilePath = path.join("src", "components", "made-with-dyad.tsx");
@@ -107,13 +111,22 @@ test("edit code edits the right file during rapid switches", async ({ po }) => {
 
   await expect
     .poll(
-      () => fs.readFileSync(path.join(appPath, firstOpenedFilePath), "utf8"),
+      () =>
+        normalizeLineEndings(
+          fs.readFileSync(path.join(appPath, firstOpenedFilePath), "utf8"),
+        ),
       { timeout: Timeout.MEDIUM },
     )
     .toEqual(firstFileEdit);
   await expect
-    .poll(() => fs.readFileSync(path.join(appPath, robotsFilePath), "utf8"), {
-      timeout: Timeout.MEDIUM,
-    })
+    .poll(
+      () =>
+        normalizeLineEndings(
+          fs.readFileSync(path.join(appPath, robotsFilePath), "utf8"),
+        ),
+      {
+        timeout: Timeout.MEDIUM,
+      },
+    )
     .toEqual(updatedRobotsFile);
 });

e2e-tests/helpers/page-objects/components/ChatActions.ts

@@ -52,9 +52,25 @@ export class ChatActions {
     }).toPass({ timeout: Timeout.SHORT });
   }
 
-  clickNewChat({ index = 0 }: { index?: number } = {}) {
-    // There is two new chat buttons...
-    return this.page.getByTestId("new-chat-button").nth(index).click();
+  async clickNewChat({ index = 0 }: { index?: number } = {}) {
+    // There are two new chat buttons.
+    const previousChatId = new URL(this.page.url()).searchParams.get("id");
+
+    await this.page.getByTestId("new-chat-button").nth(index).click();
+
+    await expect(async () => {
+      const currentChatId = new URL(this.page.url()).searchParams.get("id");
+      if (previousChatId === null) {
+        expect(currentChatId).not.toBeNull();
+      } else {
+        expect(currentChatId).not.toBe(previousChatId);
+      }
+
+      const chatInput = this.getChatInput();
+      await expect(chatInput).toBeVisible({ timeout: 1_000 });
+      const text = await chatInput.textContent({ timeout: 1_000 });
+      expect(text?.trim() ?? "").toBe("");
+    }).toPass({ timeout: Timeout.MEDIUM });
   }
 
   private getRetryButton() {
@@ -103,8 +119,8 @@ export class ChatActions {
       await chatInput.fill(prompt);
       const visiblePrompt = prompt.replace(/@app:/g, "@");
       expect(await chatInput.textContent()).toContain(visiblePrompt);
-      await expect(sendButton).toBeEnabled();
-      await sendButton.click();
+      await expect(sendButton).toBeEnabled({ timeout: 1_000 });
+      await sendButton.click({ timeout: 1_000 });
     }).toPass({ timeout: Timeout.MEDIUM });
 
     if (!skipWaitForCompletion) {

e2e-tests/helpers/page-objects/components/GitHubConnector.ts

@@ -4,6 +4,7 @@
  */
 
 import { Page, expect } from "@playwright/test";
+import { Timeout } from "../../constants";
 
 export class GitHubConnector {
   constructor(
@@ -29,6 +30,7 @@ export class GitHubConnector {
 
   async clickCreateRepoButton() {
     await this.page.getByRole("button", { name: "Create Repo" }).click();
+    await this.waitForSyncToFinish();
   }
 
   async fillCreateRepoName(name: string) {
@@ -60,9 +62,11 @@ export class GitHubConnector {
 
   async clickConnectToRepoButton() {
     await this.page.getByRole("button", { name: "Connect to repo" }).click();
+    await this.waitForSyncToFinish();
   }
 
   async snapshotConnectedRepo() {
+    await this.waitForSyncToFinish();
     await expect(
       this.page.getByTestId("github-connected-repo"),
     ).toMatchAriaSnapshot();
@@ -82,6 +86,21 @@ export class GitHubConnector {
 
   async clickSyncToGithubButton() {
     await this.page.getByRole("button", { name: "Sync to GitHub" }).click();
+    await this.waitForSyncToFinish();
+  }
+
+  async waitForSyncToFinish() {
+    await expect(
+      this.page.getByRole("button", { name: "Syncing..." }),
+    ).toBeHidden({ timeout: Timeout.LONG });
+    await expect(
+      this.page.getByRole("button", { name: "Sync to GitHub" }),
+    ).toBeEnabled({ timeout: Timeout.LONG });
+    await expect(
+      this.page
+        .getByTestId("github-connected-repo")
+        .getByText("Successfully pushed to GitHub!"),
+    ).toBeVisible({ timeout: Timeout.LONG });
   }
 
   async clickDisconnectRepoButton() {

e2e-tests/queued_message.spec.ts

 import { test, testSkipIfWindows, Timeout } from "./helpers/test_helper";
-import { expect, Locator } from "@playwright/test";
+import { expect, Locator, type Page } from "@playwright/test";
+
+async function queueMessage(page: Page, chatInput: Locator, message: string) {
+  await expect(async () => {
+    await chatInput.click();
+    await chatInput.fill(message);
+    expect(await chatInput.textContent()).toContain(message);
+    await chatInput.press("Enter");
+    await expect(page.locator("li", { hasText: message })).toBeVisible({
+      timeout: 1_000,
+    });
+  }).toPass({ timeout: Timeout.MEDIUM });
+}
 
 test.describe("queued messages", () => {
   let chatInput: Locator;
@@ -19,8 +31,7 @@ test.describe("queued messages", () => {
     await expect(chatInput).toBeVisible();
 
     // While streaming, send another message - this should be queued
-    await chatInput.fill("tc=2");
-    await chatInput.press("Enter");
+    await queueMessage(po.page, chatInput, "tc=2");
 
     // Verify the queued message indicator is visible
     // The UI shows "{count} Queued" followed by "- {status}"
@@ -55,12 +66,9 @@ test.describe("queued messages", () => {
     await expect(chatInput).toBeVisible();
 
     // Queue 3 messages while streaming
-    await chatInput.fill("tc=first");
-    await chatInput.press("Enter");
-    await chatInput.fill("tc=second");
-    await chatInput.press("Enter");
-    await chatInput.fill("tc=third");
-    await chatInput.press("Enter");
+    await queueMessage(po.page, chatInput, "tc=first");
+    await queueMessage(po.page, chatInput, "tc=second");
+    await queueMessage(po.page, chatInput, "tc=third");
 
     // Verify 3 messages are queued
     await expect(po.page.getByText("3 Queued")).toBeVisible();
@@ -120,8 +128,7 @@ test.describe("queued messages", () => {
     await expect(chatInput).toBeVisible();
 
     // While streaming, queue a second message
-    await chatInput.fill("tc=2");
-    await chatInput.press("Enter");
+    await queueMessage(po.page, chatInput, "tc=2");
 
     // Verify the queued message indicator is visible
     await expect(
@@ -193,8 +200,7 @@ testSkipIfWindows(
     await expect(po.page.getByText("logo.png")).toBeVisible();
 
     // Queue a message with both attachment and selected component
-    await chatInput.fill("queued with extras");
-    await chatInput.press("Enter");
+    await queueMessage(po.page, chatInput, "queued with extras");
 
     // After queuing, both should be cleared
     await expect(
@@ -264,8 +270,7 @@ testSkipIfWindows(
       timeout: Timeout.SHORT,
     });
 
-    await chatInput.fill("queued with component");
-    await chatInput.press("Enter");
+    await queueMessage(po.page, chatInput, "queued with component");
     await expect(po.page.getByText(/\d+ Queued/)).toBeVisible();
 
     // Edit the queued message — components should be restored

e2e-tests/visual_editing.spec.ts

-import { expect } from "@playwright/test";
+import { expect, type Page } from "@playwright/test";
 import { testSkipIfWindows, Timeout } from "./helpers/test_helper";
 const fs = require("fs");
 const path = require("path");
 
+async function saveVisualChanges(page: Page) {
+  await expect(async () => {
+    const saveButton = page.getByRole("button", { name: "Save Changes" });
+    await expect(saveButton).toBeVisible({ timeout: 1_000 });
+    await expect(saveButton).toBeEnabled({ timeout: 1_000 });
+    await saveButton.click();
+  }).toPass({ timeout: Timeout.MEDIUM });
+}
+
 testSkipIfWindows("edit style of one selected component", async ({ po }) => {
   await po.setUpDyadPro();
   await po.sendPrompt("tc=basic");
@@ -57,7 +66,7 @@ testSkipIfWindows("edit style of one selected component", async ({ po }) => {
   });
 
   // Save the changes
-  await po.page.getByRole("button", { name: "Save Changes" }).click();
+  await saveVisualChanges(po.page);
 
   // Wait for the success toast
   await po.toastNotifications.waitForToastWithText(
@@ -126,7 +135,7 @@ testSkipIfWindows("edit text of the selected component", async ({ po }) => {
   });
 
   // Save the changes
-  await po.page.getByRole("button", { name: "Save Changes" }).click();
+  await saveVisualChanges(po.page);
 
   // Wait for the success toast
   await po.toastNotifications.waitForToastWithText(
@@ -200,7 +209,7 @@ testSkipIfWindows("swap image via URL", async ({ po }) => {
   });
 
   // Save the changes
-  await po.page.getByRole("button", { name: "Save Changes" }).click();
+  await saveVisualChanges(po.page);
 
   // Wait for the success toast
   await po.toastNotifications.waitForToastWithText(

rules/e2e-testing.md

@@ -123,8 +123,10 @@ If `npm run build` fails while rebuilding native modules with `ImportError` from
 - **Keyboard navigation events (ArrowUp/ArrowDown)**: Add `await page.waitForTimeout(100)` between sequential keyboard presses to let the UI state settle. Rapid keypresses can cause race conditions in menu navigation.
 - **Navigation to tabs**: Use `await expect(link).toBeVisible({ timeout: Timeout.EXTRA_LONG })` before clicking tab links (especially in `goToAppsTab()`). Electron sidebar links can take time to render during app initialization.
 - **Confirming flakiness**: Use `PLAYWRIGHT_RETRIES=0 PLAYWRIGHT_HTML_OPEN=never npm run e2e -- e2e-tests/<spec> --repeat-each=10` to reproduce flaky tests. `PLAYWRIGHT_RETRIES=0` is critical — CI defaults to 2 retries, hiding flakiness.
+- **`expect(...).toPass()` wrappers**: Give inner Playwright actions/assertions short explicit timeouts. Default 30s click/expect timeouts can consume the whole `toPass()` budget, so the retry wrapper never actually retries.
 - **Monaco file-switch assertions**: For code-editor tests, don't stop at waiting for the editor textbox to appear. Wait until Monaco's active model URI matches the file you clicked; otherwise the test can type into a still-switching editor model and miss real file-switch races.
 - **Monaco race repros**: If a file-editor bug only appears during quick tab/file changes, alternate between the affected files several times in one test before declaring it non-reproducible. A single switch often misses save-vs-switch timing bugs that show up immediately under `--repeat-each`.
+- **GitHub sync success assertions**: Scope "Successfully pushed to GitHub!" assertions to `getByTestId("github-connected-repo")`; the same text can also appear in a toast, causing Playwright strict-mode failures.
 
 ## Real Socket Firewall E2E tests
 

src/components/chat/ChatInput.tsx

@@ -26,7 +26,7 @@ import { useTranslation } from "react-i18next";
 import { useSettings } from "@/hooks/useSettings";
 import { ipc } from "@/ipc/types";
 import {
-  chatInputValueAtom,
+  chatInputValuesByIdAtom,
   chatMessagesByIdAtom,
   selectedChatIdAtom,
   pendingAgentConsentsAtom,
@@ -113,7 +113,23 @@ const showTokenBarAtom = atom(false);
 export function ChatInput({ chatId }: { chatId?: number }) {
   const { t } = useTranslation("chat");
   const posthog = usePostHog();
-  const [inputValue, setInputValue] = useAtom(chatInputValueAtom);
+  const inputValuesById = useAtomValue(chatInputValuesByIdAtom);
+  const setInputValuesById = useSetAtom(chatInputValuesByIdAtom);
+  const inputValue = chatId ? (inputValuesById.get(chatId) ?? "") : "";
+  const setInputValue = useCallback(
+    (newValue: string | ((prev: string) => string)) => {
+      if (!chatId) return;
+
+      setInputValuesById((currentMap) => {
+        const prev = currentMap.get(chatId) ?? "";
+        const next = typeof newValue === "function" ? newValue(prev) : newValue;
+        const newMap = new Map(currentMap);
+        newMap.set(chatId, next);
+        return newMap;
+      });
+    },
+    [chatId, setInputValuesById],
+  );
   const { settings } = useSettings();
   const {
     selectedMode: chatMode,

src/ipc/handlers/chat_stream_handlers.ts

@@ -1131,7 +1131,7 @@ This conversation includes one or more image attachments. When the user uploads
             system: systemPromptOverride,
             tools,
             messages: chatMessages.filter((m) => m.content),
-            onFinish: (response) => {
+            onFinish: async (response) => {
               const totalTokens = response.usage?.totalTokens;
 
               if (typeof totalTokens === "number") {
@@ -1140,7 +1140,7 @@ This conversation includes one or more image attachments. When the user uploads
                 maxTokensUsed = Math.max(maxTokensUsed ?? 0, totalTokens);
 
                 // Persist the aggregated token usage on the placeholder assistant message
-                void db
+                await db
                   .update(messages)
                   .set({ maxTokensUsed: maxTokensUsed })
                   .where(eq(messages.id, placeholderAssistantMessage.id))