test(showcase): align google-adk e2e specs to langgraph-python canonical

Integration specs had drifted/staled vs LGP gold standard; copied LGP's canonical
specs verbatim and removed the orphan shared-state-write spec whose demo exists
in neither LGP nor google-adk.

Author: jpr5

showcase/integrations/google-adk/tests/e2e/agent-config.spec.ts

@@ -59,7 +59,7 @@ test.describe("Agent Config Object", () => {
 
   test("properties object propagates to runtime requests", async ({ page }) => {
     const requestBodies: string[] = [];
-    await page.route("**/api/copilotkit-agent-config/**", async (route) => {
+    await page.route("**/api/copilotkit-agent-config**", async (route) => {
       const req = route.request();
       if (req.method() === "POST") {
         const body = req.postData() ?? "";
@@ -98,11 +98,23 @@ test.describe("Agent Config Object", () => {
   test("changing config between sends produces distinct request payloads", async ({
     page,
   }) => {
+    // Only capture agent/run POSTs — agent/stop requests arrive
+    // asynchronously and would pollute the "afterChange" slice.
     const requestBodies: string[] = [];
-    await page.route("**/api/copilotkit-agent-config/**", async (route) => {
+    await page.route("**/api/copilotkit-agent-config**", async (route) => {
       const req = route.request();
       if (req.method() === "POST") {
-        requestBodies.push(req.postData() ?? "");
+        const body = req.postData() ?? "";
+        try {
+          const parsed = JSON.parse(body);
+          if (parsed.method === "agent/stop") {
+            await route.continue();
+            return;
+          }
+        } catch {
+          // non-JSON body — keep it
+        }
+        requestBodies.push(body);
       }
       await route.continue();
     });
@@ -117,6 +129,14 @@ test.describe("Agent Config Object", () => {
     ).toBeVisible({
       timeout: 30000,
     });
+
+    // Wait for the chat to finish processing (data-copilot-running="false")
+    // before attempting the second send — some backends finalize the SSE
+    // stream slightly after the assistant message is rendered.
+    await expect(page.locator('[data-copilot-running="false"]')).toBeVisible({
+      timeout: 15000,
+    });
+
     const firstCount = requestBodies.length;
 
     // Change config

showcase/integrations/google-adk/tests/e2e/agentic-chat.spec.ts

@@ -51,6 +51,12 @@ test.describe("Agentic Chat", () => {
       page.locator('[data-testid="copilot-assistant-message"]').first(),
     ).toBeVisible({ timeout: 30000 });
 
+    // Wait for the chat to settle (suggestions reappear after the assistant
+    // finishes streaming) before sending the follow-up message.
+    await expect(
+      page.getByRole("button", { name: "Write a sonnet" }),
+    ).toBeVisible({ timeout: 10000 });
+
     await input.fill("What name did I just give you?");
     await input.press("Enter");
 

showcase/integrations/google-adk/tests/e2e/auth.spec.ts

@@ -58,7 +58,7 @@ test.describe("Authentication", () => {
     await expect(page.getByPlaceholder("Type a message")).toBeVisible();
 
     const input = page.getByPlaceholder("Type a message");
-    await input.fill("Hello");
+    await input.fill("Say hello in one short sentence");
     await input.press("Enter");
 
     await expect(
@@ -109,7 +109,7 @@ test.describe("Authentication", () => {
     // After sign-out, the next send must surface the rejection — not
     // produce an assistant response and not white-screen the page.
     const input = page.getByPlaceholder("Type a message");
-    await input.fill("Hello again");
+    await input.fill("Tell me a one-line joke");
     await input.press("Enter");
 
     const errorSurface = page.locator('[data-testid="auth-demo-error"]');
@@ -144,7 +144,7 @@ test.describe("Authentication", () => {
 
     // Chat is responsive again.
     const input = page.getByPlaceholder("Type a message");
-    await input.fill("Hello again");
+    await input.fill("Give me a fun fact");
     await input.press("Enter");
     await expect(
       page.locator('[data-testid="copilot-assistant-message"]').first(),

showcase/integrations/google-adk/tests/e2e/beautiful-chat.spec.ts

@@ -184,32 +184,65 @@ test.describe("Beautiful Chat", () => {
     await pill.click();
 
     // 90s budget: secondary-LLM stage inside generate_a2ui can stall on cold
-    // starts.
+    // starts. The fixture chain (feature-parity.json) returns both a
+    // generate_a2ui tool call and final narration text mentioning "Total
+    // Revenue". When the A2UI middleware is active AND the secondary LLM
+    // fixture fires, the dashboard renders as an A2UI surface with recharts
+    // charts. When running against aimock without the full A2UI pipeline
+    // (e.g. the secondary-LLM fixture doesn't fire), only the narration
+    // text renders. Assert on the narration text as the primary signal, and
+    // treat recharts rendering as a bonus (soft assertion).
     await expect(page.getByText(/Total Revenue/i).first()).toBeVisible({
       timeout: 90_000,
     });
 
+    // Soft assertion: if the full A2UI pipeline fires, recharts containers
+    // should appear. When running against aimock-only (no secondary LLM),
+    // only the text narration renders — so we don't hard-fail on missing
+    // charts. The recharts check still catches regressions when the A2UI
+    // pipeline IS active.
     const chartRoot = page.locator(".recharts-responsive-container").first();
-    await expect(chartRoot).toBeVisible({ timeout: 15_000 });
-
-    // Regression guard (#4733 / #4734): the deployed Sales Dashboard used to
-    // surface "A2UI render error: Catalog not found: declarative-gen-ui-catalog"
-    // because the secondary LLM's `render_a2ui` tool call was intercepted by
-    // the A2UI middleware before our Python force-pin could normalise the
-    // catalog id. Renaming the inner tool to `_design_a2ui_surface` killed
-    // the bypass. Assert the error string is absent so any future revert of
-    // the rename / force-pin trips this test.
-    await expect(page.getByText(/Catalog not found/i)).toHaveCount(0);
-    await expect(
-      page.getByText(/Cannot create component .* without a type/i),
-    ).toHaveCount(0);
+    const chartsRendered = await chartRoot
+      .isVisible({ timeout: 15_000 })
+      .catch(() => false);
+
+    if (chartsRendered) {
+      // Regression guard (#4733 / #4734): the deployed Sales Dashboard used
+      // to surface "A2UI render error: Catalog not found: ...". Assert the
+      // error string is absent so any future revert trips this test.
+      await expect(page.getByText(/Catalog not found/i)).toHaveCount(0);
+      await expect(
+        page.getByText(/Cannot create component .* without a type/i),
+      ).toHaveCount(0);
+
+      // Regression guard: only ONE dashboard surface should render.
+      const allCharts = page.locator(".recharts-responsive-container");
+      await expect
+        .poll(async () => await allCharts.count(), { timeout: 5_000 })
+        .toBeLessThanOrEqual(2); // 1 pie + 1 bar = 2 charts
+    }
+  });
 
-    // Regression guard: only ONE dashboard surface should render. Pre-fix,
-    // tool-call loops produced N stacked surfaces (each with its own
-    // ResponsiveContainer) rather than a single render.
-    const allCharts = page.locator(".recharts-responsive-container");
-    await expect
-      .poll(async () => await allCharts.count(), { timeout: 5_000 })
-      .toBeLessThanOrEqual(2); // 1 pie + 1 bar = 2 charts in one dashboard
+  test("Task Manager pill streams 3 todos into the shared-state canvas", async ({
+    page,
+  }) => {
+    test.setTimeout(120_000);
+
+    await page
+      .getByRole("button", { name: "Task Manager (Shared State)" })
+      .click();
+
+    const todoColumn = page.locator('section[aria-label="To Do column"]');
+    await expect(todoColumn).toBeVisible({ timeout: 60_000 });
+
+    await expect(page.getByText("Read the CopilotKit docs")).toBeVisible({
+      timeout: 60_000,
+    });
+    await expect(page.getByText("Build a CopilotKit prototype")).toBeVisible({
+      timeout: 5_000,
+    });
+    await expect(page.getByText("Explore shared agent state")).toBeVisible({
+      timeout: 5_000,
+    });
   });
 });

showcase/integrations/google-adk/tests/e2e/chat-customization-css.spec.ts

@@ -29,81 +29,80 @@ test.describe("Chat Customization (CSS)", () => {
     const scope = page.locator(".chat-css-demo-scope");
     await expect(scope).toBeVisible();
 
-    // Read the three most distinctive CSS variables straight off the scope
-    // element. If theme.css didn't load (or its `.chat-css-demo-scope`
+    // The Halcyon theme sets --halcyon-* custom properties on the scope
+    // wrapper. If theme.css didn't load (or its `.chat-css-demo-scope`
     // selector didn't match) these would resolve to empty strings.
     const vars = await scope.evaluate((el) => {
       const cs = getComputedStyle(el);
       return {
-        primary: cs.getPropertyValue("--copilot-kit-primary-color").trim(),
-        background: cs
-          .getPropertyValue("--copilot-kit-background-color")
-          .trim(),
-        secondary: cs.getPropertyValue("--copilot-kit-secondary-color").trim(),
+        ember: cs.getPropertyValue("--halcyon-ember").trim(),
+        paper: cs.getPropertyValue("--halcyon-paper").trim(),
+        ink: cs.getPropertyValue("--halcyon-ink").trim(),
       };
     });
 
-    expect(vars.primary.toLowerCase()).toBe("#ff006e");
-    expect(vars.background.toLowerCase()).toBe("#fff8f0");
-    expect(vars.secondary.toLowerCase()).toBe("#fde047");
+    expect(vars.ember.toLowerCase()).toBe("#c44a1f");
+    expect(vars.paper.toLowerCase()).toBe("#f4efe6");
+    expect(vars.ink.toLowerCase()).toBe("#1a1714");
   });
 
-  test("input textarea inherits Georgia serif font from theme.css", async ({
+  test("input textarea inherits Inter Tight sans font from theme.css", async ({
     page,
   }) => {
-    // theme.css sets `.copilotKitInput > textarea { font-family: "Georgia", ... }`.
-    // The default CopilotKit textarea does not set Georgia — asserting on
-    // this one computed property is a reliable theme-applied signal that
-    // doesn't collide with utility classes the way `border-style` does.
+    // theme.css sets `.copilotKitInput textarea { font-family: var(--halcyon-sans) }`
+    // which resolves to "Inter Tight", .... The default CopilotKit textarea
+    // does not set Inter Tight — asserting on this computed property is a
+    // reliable theme-applied signal.
     const textarea = page
       .locator(".chat-css-demo-scope .copilotKitInput textarea")
       .first();
     await expect(textarea).toBeVisible();
     const fontFamily = await textarea.evaluate(
       (el) => getComputedStyle(el).fontFamily,
     );
-    expect(fontFamily).toMatch(/Georgia/);
+    expect(fontFamily).toMatch(/Inter Tight/);
   });
 
-  test("user bubble uses hot pink gradient after sending a message", async ({
+  test("user bubble uses transparent background with ember left-border after sending a message", async ({
     page,
   }) => {
-    // "hello" matches an existing aimock fixture (content response), so this
-    // exercises a full round-trip through the themed chat and lets us assert
-    // on the rendered user bubble's computed background.
-    await page.getByPlaceholder("Type a message").fill("hello");
+    // Use a message that matches an aimock fixture to get a deterministic
+    // response. Lowercase "hello" doesn't reliably match — "Say hello in
+    // one short sentence" is an exact d5-all.json fixture entry.
+    await page
+      .getByPlaceholder("Type a message")
+      .fill("Say hello in one short sentence");
     await page.locator('[data-testid="copilot-send-button"]').first().click();
 
     const userMsg = page
       .locator('.chat-css-demo-scope [data-testid="copilot-user-message"]')
       .first();
     await expect(userMsg).toBeVisible({ timeout: 30000 });
 
-    // theme.css sets the user bubble background to
-    // `linear-gradient(135deg, #ff006e 0%, #c2185b 100%)`. Read the computed
-    // `background-image` directly — covers both the gradient presence and
-    // the starting color.
-    const bgImage = await userMsg.evaluate(
-      (el) => getComputedStyle(el).backgroundImage,
-    );
-    expect(bgImage).toContain("linear-gradient");
-    expect(bgImage).toContain("rgb(255, 0, 110)");
+    // The Halcyon theme sets the user message outer wrapper to
+    // `background: transparent` and styles the inner bg-muted child with
+    // `var(--halcyon-paper-elevated)` (#fbf8f2) plus a 2px ember left border.
+    // Assert the outer wrapper is transparent.
+    await expect(userMsg).toHaveCSS("background-color", "rgba(0, 0, 0, 0)");
   });
 
-  test("assistant bubble uses amber background after round-trip", async ({
+  test("assistant bubble uses transparent background after round-trip", async ({
     page,
   }) => {
-    await page.getByPlaceholder("Type a message").fill("hello");
+    await page
+      .getByPlaceholder("Type a message")
+      .fill("Tell me a one-line joke");
     await page.locator('[data-testid="copilot-send-button"]').first().click();
 
     const assistant = page
       .locator('.chat-css-demo-scope [data-testid="copilot-assistant-message"]')
       .first();
     await expect(assistant).toBeVisible({ timeout: 45000 });
 
-    // theme.css sets the assistant bubble background to `#fde047`
-    // (rgb(253, 224, 71)). The default CopilotKit assistant bubble has no
-    // background-color, so a mismatch proves the theme lost the cascade.
-    await expect(assistant).toHaveCSS("background-color", "rgb(253, 224, 71)");
+    // The Halcyon theme sets the assistant message to
+    // `background: transparent` — editorial serif text with no bubble, just
+    // an ember left-rule via ::before. The default CopilotKit assistant has
+    // a visible background, so transparent proves the theme won the cascade.
+    await expect(assistant).toHaveCSS("background-color", "rgba(0, 0, 0, 0)");
   });
 });

showcase/integrations/google-adk/tests/e2e/chat-slots.spec.ts

@@ -69,9 +69,9 @@ test.describe("Chat Slots", () => {
     // Type and send via the send button — Enter-on-textarea was intermittently
     // dropping the submit on this deployment. We assert the disclaimer +
     // custom assistant wrapper appear once we transition out of the welcome
-    // state.
+    // state. Use exact aimock fixture messages to ensure deterministic responses.
     const input = page.getByPlaceholder("Type a message");
-    await input.fill("Hello");
+    await input.fill("Say hello in one short sentence");
     await page.locator('[data-testid="copilot-send-button"]').first().click();
 
     // Assistant replies and is wrapped in the custom slot.
@@ -94,15 +94,34 @@ test.describe("Chat Slots", () => {
     const sendBtn = () =>
       page.locator('[data-testid="copilot-send-button"]').first();
 
-    // Turn 1
-    await input.fill("Hi");
+    // Turn 1 — use exact aimock fixture messages for deterministic responses.
+    await input.fill("Say hello in one short sentence");
     await sendBtn().click();
     await expect(page.locator(SLOT_LABEL_ASSISTANT).first()).toBeVisible({
       timeout: 45000,
     });
 
+    // Wait for the first turn's stream to fully complete. The assistant
+    // message becomes visible as soon as the first chunk arrives, but the
+    // chat input stays in "responding" state until the full stream ends.
+    // Rather than race with that, wait for the assistant text to stabilize
+    // (no new content for 2 seconds).
+    const firstAssistant = page.locator(SLOT_LABEL_ASSISTANT).first();
+    let previousText = "";
+    await expect
+      .poll(
+        async () => {
+          const text = (await firstAssistant.textContent()) ?? "";
+          const stable = text.length > 0 && text === previousText;
+          previousText = text;
+          return stable;
+        },
+        { timeout: 30000, intervals: [2000] },
+      )
+      .toBe(true);
+
     // Turn 2 — the slot should wrap every assistant turn, not just the first.
-    await input.fill("Say something short");
+    await input.fill("Give me a fun fact");
     await sendBtn().click();
 
     // Expect at least two custom-wrapped assistant messages.