test(ci): exercise full validation suite

.github/workflows/ci-full.yml

@@ -287,13 +287,21 @@ jobs:
           mkdir -p "$OPENHUMAN_WORKSPACE"
           bash scripts/ci-cancel-aware.sh bash app/scripts/e2e-web-session.sh
 
+      - name: Pack Playwright E2E failure artifacts
+        if: failure()
+        run: |
+          mkdir -p .ci/artifacts
+          tar -czf .ci/artifacts/openhuman-playwright-failure-logs.tar.gz \
+            -C "$OPENHUMAN_WORKSPACE" .
+        env:
+          OPENHUMAN_WORKSPACE: ${{ runner.temp }}/openhuman-playwright-workspace
+
       - name: Upload Playwright E2E failure artifacts
         if: failure()
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7
         with:
           name: e2e-playwright-failure-logs-${{ github.run_id }}
-          path: |
-            ${{ runner.temp }}/openhuman-playwright-workspace/**
+          path: .ci/artifacts/openhuman-playwright-failure-logs.tar.gz
           retention-days: 7
           if-no-files-found: ignore
 

.github/workflows/e2e-playwright.yml

@@ -26,7 +26,10 @@ jobs:
     runs-on: ubuntu-22.04
     container:
       image: ghcr.io/tinyhumansai/openhuman_ci:latest
-    timeout-minutes: 30
+    # The complete serial web suite currently takes about 45 minutes on the
+    # shared runner; keep the standalone diagnostic workflow aligned with the
+    # 90-minute budget used by CI Full.
+    timeout-minutes: 90
     steps:
       - name: Checkout code
         uses: actions/checkout@v7
@@ -73,12 +76,20 @@ jobs:
           mkdir -p "$OPENHUMAN_WORKSPACE"
           bash scripts/ci-cancel-aware.sh bash app/scripts/e2e-web-session.sh
 
+      - name: Pack Playwright E2E failure artifacts
+        if: failure()
+        run: |
+          mkdir -p .ci/artifacts
+          tar -czf .ci/artifacts/openhuman-playwright-failure-logs.tar.gz \
+            -C "$OPENHUMAN_WORKSPACE" .
+        env:
+          OPENHUMAN_WORKSPACE: ${{ runner.temp }}/openhuman-playwright-workspace
+
       - name: Upload Playwright E2E failure artifacts
         if: failure()
         uses: actions/upload-artifact@v7
         with:
           name: e2e-playwright-failure-logs-${{ github.run_id }}
-          path: |
-            ${{ runner.temp }}/openhuman-playwright-workspace/**
+          path: .ci/artifacts/openhuman-playwright-failure-logs.tar.gz
           retention-days: 7
           if-no-files-found: ignore

.github/workflows/e2e-reusable.yml

@@ -291,7 +291,8 @@ jobs:
           - { name: provider-web, suites: "provider-web" }
           - { name: webhooks, suites: "webhooks" }
           - { name: connectors, suites: "connectors" }
-          - { name: commerce, suites: "payments,settings" }
+          - { name: payments, suites: "payments" }
+          - { name: settings, suites: "settings" }
     steps:
       - name: Checkout code
         uses: actions/checkout@v7
@@ -765,7 +766,8 @@ jobs:
           - { name: provider-web, suites: "provider-web" }
           - { name: webhooks, suites: "webhooks" }
           - { name: connectors, suites: "connectors" }
-          - { name: commerce, suites: "payments,settings" }
+          - { name: payments, suites: "payments" }
+          - { name: settings, suites: "settings" }
     steps:
       - name: Checkout code
         uses: actions/checkout@v7
@@ -976,7 +978,8 @@ jobs:
           - { name: provider-web, suites: "provider-web" }
           - { name: webhooks, suites: "webhooks" }
           - { name: connectors, suites: "connectors" }
-          - { name: commerce, suites: "payments,settings" }
+          - { name: payments, suites: "payments" }
+          - { name: settings, suites: "settings" }
     steps:
       - name: Checkout code
         uses: actions/checkout@v7

app/scripts/e2e-run-shards.sh

@@ -18,7 +18,8 @@
 #   chat         = chat, skills, journeys
 #   integrations = providers, webhooks, notifications
 #   connectors   = connectors
-#   commerce     = payments, settings
+#   payments     = payments
+#   settings     = settings
 #
 set -uo pipefail
 
@@ -32,7 +33,8 @@ SHARDS=(
   "providers:providers,notifications"
   "webhooks:webhooks"
   "connectors:connectors"
-  "commerce:payments,settings"
+  "payments:payments"
+  "settings:settings"
 )
 
 # Allow filtering: `bash e2e-run-shards.sh foundation chat`

app/test/core-rpc-node.test.ts

@@ -1,4 +1,4 @@
-import { describe, expect, it } from 'vitest';
+import { afterEach, describe, expect, it, vi } from 'vitest';
 
 import { formatRpcCallFailure } from './e2e/helpers/core-rpc-node';
 
@@ -15,3 +15,54 @@ describe('formatRpcCallFailure', () => {
     );
   });
 });
+
+describe('callOpenhumanRpcNode', () => {
+  afterEach(() => {
+    vi.unstubAllGlobals();
+    vi.resetModules();
+  });
+
+  it('rediscovers the core when the cached listener disappears after a reset', async () => {
+    const requestedUrls: string[] = [];
+    let firstListenerAlive = true;
+
+    vi.stubGlobal(
+      'fetch',
+      vi.fn(async (input: string | URL | Request, init?: RequestInit) => {
+        const url = String(input);
+        requestedUrls.push(url);
+        const body = JSON.parse(String(init?.body)) as { method: string };
+
+        if (url.includes(':7788/')) {
+          if (body.method === 'core.ping' && firstListenerAlive) {
+            return new Response('', { status: 401 });
+          }
+          if (body.method === 'openhuman.first_call' && firstListenerAlive) {
+            return Response.json({ result: 'first' });
+          }
+          throw new TypeError('fetch failed');
+        }
+
+        if (url.includes(':7789/')) {
+          if (body.method === 'core.ping') return new Response('', { status: 401 });
+          return Response.json({ result: 'replacement' });
+        }
+
+        throw new TypeError('fetch failed');
+      })
+    );
+
+    const { callOpenhumanRpcNode } = await import('./e2e/helpers/core-rpc-node');
+    await expect(callOpenhumanRpcNode('openhuman.first_call')).resolves.toMatchObject({
+      ok: true,
+      result: 'first',
+    });
+
+    firstListenerAlive = false;
+    await expect(callOpenhumanRpcNode('openhuman.after_reset')).resolves.toMatchObject({
+      ok: true,
+      result: 'replacement',
+    });
+    expect(requestedUrls.some(url => url.includes(':7789/rpc'))).toBe(true);
+  });
+});

app/test/e2e/helpers/chat-harness.ts

@@ -78,42 +78,40 @@ export async function chatMounted(): Promise<boolean> {
 /** Type into the chat composer through WebDriver so React's controlled
  *  input state and the DOM stay in sync. */
 export async function typeIntoComposer(text: string): Promise<void> {
-  const composer = await browser.$(COMPOSER_SELECTOR);
-  await composer.waitForDisplayed({ timeout: 10_000 });
-  await composer.waitForEnabled({ timeout: 10_000 });
+  let actual = '';
+  for (let attempt = 1; attempt <= 3; attempt += 1) {
+    // Creating a thread can replace the controlled textarea after the selected
+    // thread id changes. Resolve it afresh on every attempt so a late React
+    // commit cannot leave WebDriver typing into a detached element.
+    const composer = await browser.$(COMPOSER_SELECTOR);
+    await composer.waitForDisplayed({ timeout: 10_000 });
+    await composer.waitForEnabled({ timeout: 10_000 });
 
-  // Step 1: Focus via JS — avoids the coordinate-based click that gets
-  // intercepted by AppUpdatePrompt (z-[9998], fixed bottom-4 right-4).
-  // We also select-all any existing text so the subsequent delete clears it.
-  const focused = await browser.execute((sel: string) => {
-    const el = document.querySelector(sel) as HTMLTextAreaElement | null;
-    if (!el) return false;
-    el.focus();
-    el.select();
-    return true;
-  }, COMPOSER_SELECTOR);
-  if (!focused) {
-    throw new Error('typeIntoComposer: textarea not found');
-  }
+    // Focus via JS — avoids the coordinate-based click that gets intercepted
+    // by AppUpdatePrompt. Select any partial value before deleting it.
+    const focused = await browser.execute((sel: string) => {
+      const el = document.querySelector(sel) as HTMLTextAreaElement | null;
+      if (!el) return false;
+      el.focus();
+      el.select();
+      return true;
+    }, COMPOSER_SELECTOR);
+    if (!focused) continue;
 
-  // Step 2: Clear existing content.  el.select() inside browser.execute already
-  // selected all text; browser.keys('Delete') now removes the selection so
-  // React's controlled state sees an empty value before we start typing.
-  await browser.pause(80);
-  await browser.keys('Delete');
-  await browser.pause(80);
+    await browser.pause(80);
+    await browser.keys('Delete');
+    await browser.pause(80);
 
-  // Step 3: Type the text using real OS-level keyboard events (browser.keys).
-  // Unlike synthetic DOM events dispatched via browser.execute(), these go
-  // through Chromium's normal input pipeline, triggering React's onChange
-  // on the controlled textarea and correctly updating `inputValue` state so
-  // the send button becomes enabled.
-  await browser.keys(text.split(''));
+    // Real keyboard events keep React's controlled state and the DOM in sync.
+    await browser.keys(text.split(''));
+    await browser.pause(200);
+    actual = String(await composer.getValue());
+    if (actual === text) return;
+  }
 
-  await browser.waitUntil(async () => (await composer.getValue()) === text, {
-    timeout: 5_000,
-    timeoutMsg: 'chat composer did not receive typed text',
-  });
+  throw new Error(
+    `chat composer did not receive typed text after 3 attempts (actual length ${actual.length}, expected ${text.length})`
+  );
 }
 
 /** Click the chat composer's send button. Returns `false` if the

app/test/e2e/helpers/core-rpc-node.ts

@@ -90,7 +90,13 @@ function coreHost(): string {
   return (process.env.OPENHUMAN_CORE_HOST || '127.0.0.1').trim() || '127.0.0.1';
 }
 
-/** Ports to try when OPENHUMAN_CORE_PORT is unset (matches typical dev sidecar range). */
+/** Ports to try when OPENHUMAN_CORE_PORT is unset.
+ *
+ * Keep this exactly aligned with connectivity::rpc's desktop fallback range.
+ * A data reset restarts the embedded core; on Windows the preferred socket can
+ * remain unavailable briefly, so the replacement listener may bind as high as
+ * 7798. Stopping at 7793 makes every later RPC test wait out the full probe
+ * deadline even though the restarted core is healthy. */
 function defaultPortProbeList(): number[] {
   const raw = process.env.OPENHUMAN_CORE_PORT?.trim();
   if (raw) {
@@ -100,7 +106,7 @@ function defaultPortProbeList(): number[] {
     }
   }
   const ports: number[] = [];
-  for (let port = 7788; port <= 7793; port += 1) ports.push(port);
+  for (let port = 7788; port <= 7798; port += 1) ports.push(port);
   return ports;
 }
 
@@ -113,6 +119,9 @@ async function tryPingRpc(url: string): Promise<boolean> {
       method: 'POST',
       headers: buildHeaders(false),
       body: JSON.stringify({ jsonrpc: '2.0', id: 1, method: 'core.ping', params: {} }),
+      // A recently stopped listener can take several seconds to reject on
+      // Windows. Keep discovery inside resetApp's eight-second RPC budget.
+      signal: AbortSignal.timeout(750),
     });
     // 401 means "endpoint exists, auth required" — that's a positive match
     // for the core RPC URL; the real call will retry with auth attached.
@@ -130,7 +139,10 @@ async function tryPingRpc(url: string): Promise<boolean> {
  * `OPENHUMAN_CORE_HOST` + `OPENHUMAN_CORE_PORT`, then probe host:port until core.ping succeeds.
  */
 export async function resolveCoreRpcUrl(): Promise<string> {
-  if (cachedRpcUrl) return cachedRpcUrl;
+  if (cachedRpcUrl) {
+    if (await tryPingRpc(cachedRpcUrl)) return cachedRpcUrl;
+    cachedRpcUrl = null;
+  }
 
   const env = process.env.OPENHUMAN_CORE_RPC_URL?.trim();
   if (env) {
@@ -162,33 +174,43 @@ export async function callOpenhumanRpcNode<T = unknown>(
   method: string,
   params: Record<string, unknown> = {}
 ): Promise<RpcCallResult<T>> {
-  try {
-    const rpcUrl = await resolveCoreRpcUrl();
-    const id = Math.floor(Math.random() * 1e9);
-    const res = await fetch(rpcUrl, {
-      method: 'POST',
-      headers: buildHeaders(),
-      body: JSON.stringify({ jsonrpc: '2.0', id, method, params }),
-    });
-    const text = await res.text();
-    let json: { error?: { message?: string }; result?: T };
+  for (let attempt = 0; attempt < 2; attempt += 1) {
     try {
-      json = JSON.parse(text) as typeof json;
-    } catch {
-      return {
-        ok: false,
-        httpStatus: res.status,
-        error: `Invalid JSON (${res.status}): ${text.slice(0, 240)}`,
-      };
-    }
-    if (!res.ok) {
-      return { ok: false, httpStatus: res.status, error: text.slice(0, 500) };
-    }
-    if (json.error) {
-      return { ok: false, error: json.error.message || JSON.stringify(json.error) };
+      const rpcUrl = await resolveCoreRpcUrl();
+      const id = Math.floor(Math.random() * 1e9);
+      const res = await fetch(rpcUrl, {
+        method: 'POST',
+        headers: buildHeaders(),
+        body: JSON.stringify({ jsonrpc: '2.0', id, method, params }),
+      });
+      const text = await res.text();
+      let json: { error?: { message?: string }; result?: T };
+      try {
+        json = JSON.parse(text) as typeof json;
+      } catch {
+        return {
+          ok: false,
+          httpStatus: res.status,
+          error: `Invalid JSON (${res.status}): ${text.slice(0, 240)}`,
+        };
+      }
+      if (!res.ok) {
+        return { ok: false, httpStatus: res.status, error: text.slice(0, 500) };
+      }
+      if (json.error) {
+        return { ok: false, error: json.error.message || JSON.stringify(json.error) };
+      }
+      return { ok: true, result: json.result };
+    } catch (e) {
+      // A data reset can restart the embedded core on another fallback port.
+      // Discard a cached listener after a transport failure and discover the
+      // replacement once before surfacing the error to the spec.
+      cachedRpcUrl = null;
+      if (attempt === 1) {
+        return { ok: false, error: e instanceof Error ? e.message : String(e) };
+      }
     }
-    return { ok: true, result: json.result };
-  } catch (e) {
-    return { ok: false, error: e instanceof Error ? e.message : String(e) };
   }
+
+  return { ok: false, error: 'Core JSON-RPC retry exhausted' };
 }

app/test/e2e/specs/chat-multi-tool-round.spec.ts

@@ -140,6 +140,7 @@ describe('Chat multi-tool round', () => {
 
     // Watch for file_read to appear in the timeline.
     let sawFileRead = false;
+    let sawFinal = false;
     const deadline = Date.now() + 45_000;
     while (Date.now() < deadline) {
       const snap = await getToolTimeline(threadId);
@@ -149,14 +150,15 @@ describe('Chat multi-tool round', () => {
         break;
       }
       if (await textExists(CANARY_FINAL)) {
+        sawFinal = true;
         console.log(`${LOG_PREFIX} T2.1: final answer arrived (tools may have already cycled)`);
         break;
       }
       await browser.pause(200);
     }
 
     const finalArrived = await textExists(CANARY_FINAL);
-    expect(sawFileRead || finalArrived).toBe(true);
+    expect(sawFileRead || sawFinal || finalArrived).toBe(true);
     console.log(`${LOG_PREFIX} T2.1: passed`);
   });