From 4e9402a441b2ccfb8a3af69690f601469867fd22 Mon Sep 17 00:00:00 2001
From: Brennan Benson <79079362+brennanb2025@users.noreply.github.com>
Date: Sun, 8 Nov 2026 22:48:02 -0800
Subject: [PATCH] Use the history page's itemId in the 4a harness and name what
 a visible-state wait timed out on

---
 .../native-chat/native-chat-4a-ops-reads.ts   |  2 +-
 .../native-chat/native-chat-4a-probes.ts      | 34 ++++++++++++++---
 .../native-chat/native-chat-4a-session.ts     | 37 ++++++++++++-------
 3 files changed, 52 insertions(+), 21 deletions(-)

diff --git a/config/scripts/performance/native-chat/native-chat-4a-ops-reads.ts b/config/scripts/performance/native-chat/native-chat-4a-ops-reads.ts
index 47b3bb96bb3..8846cdf7b38 100644
--- a/config/scripts/performance/native-chat/native-chat-4a-ops-reads.ts
+++ b/config/scripts/performance/native-chat/native-chat-4a-ops-reads.ts
@@ -94,7 +94,7 @@ export async function historyRows(ctx: RunContext): Promise<void> {
       continue
     }
     await sampled(ctx, row, n, () =>
-      read(ctx, 'agentSession.item', { sessionId, itemId: entry.item.id })
+      read(ctx, 'agentSession.item', { sessionId, itemId: entry.item.itemId })
     )
   }
 }
diff --git a/config/scripts/performance/native-chat/native-chat-4a-probes.ts b/config/scripts/performance/native-chat/native-chat-4a-probes.ts
index cff44fe4ca9..48bdd599775 100644
--- a/config/scripts/performance/native-chat/native-chat-4a-probes.ts
+++ b/config/scripts/performance/native-chat/native-chat-4a-probes.ts
@@ -94,15 +94,37 @@ export async function markerSeenAt(
   marker: string,
   timeoutMs = 120_000
 ): Promise<number> {
-  const handle = await page.waitForFunction(
-    (marker) =>
-      (window as unknown as { __4a: { seen: Map<string, number> } }).__4a.seen.get(marker) ?? false,
-    marker,
-    { timeout: timeoutMs, polling: 16 }
-  )
+  const handle = await page
+    .waitForFunction(
+      (marker) =>
+        (window as unknown as { __4a: { seen: Map<string, number> } }).__4a.seen.get(marker) ??
+        false,
+      marker,
+      { timeout: timeoutMs, polling: 16 }
+    )
+    .catch(async (error: unknown) => {
+      throw new Error(
+        `${marker} not visible in ${timeoutMs} ms (${await visibleChatState(page)})`,
+        {
+          cause: error
+        }
+      )
+    })
   return (await handle.jsonValue()) as number
 }
 
+/** One line on the chat the user sees, for timeout messages. */
+export function visibleChatState(page: Page): Promise<string> {
+  return page
+    .evaluate(() => {
+      const roots = [...document.querySelectorAll('[data-native-chat-root="true"]')]
+      const root = roots.find((candidate) => candidate.checkVisibility())
+      const text = root?.textContent ?? ''
+      return `${roots.length} roots, visible=${Boolean(root)}, working=${root?.getAttribute('data-native-chat-working')}, tail=${JSON.stringify(text.slice(-160))}`
+    })
+    .catch((error: unknown) => `state unreadable: ${String(error)}`)
+}
+
 /** Gaps between visible transcript updates inside [from, to], the live-update cadence a reader sees. */
 export async function mutationGaps(page: Page, from: number, to: number): Promise<number[]> {
   return page.evaluate(
diff --git a/config/scripts/performance/native-chat/native-chat-4a-session.ts b/config/scripts/performance/native-chat/native-chat-4a-session.ts
index 522a8741524..c886b2d7bb2 100644
--- a/config/scripts/performance/native-chat/native-chat-4a-session.ts
+++ b/config/scripts/performance/native-chat/native-chat-4a-session.ts
@@ -4,9 +4,9 @@ import type {} from '../../../../tests/e2e/helpers/runtime-types'
 import { randomUUID } from 'node:crypto'
 import type { Page } from '@stablyai/playwright-test'
 import { operationId, type Arm } from './native-chat-4a-arm'
-import { rpc } from './native-chat-4a-probes'
+import { rpc, visibleChatState } from './native-chat-4a-probes'
 
-type HistoryItem = { id: string; body?: { kind?: string; role?: string; state?: string } }
+type HistoryItem = { itemId: string; body?: { kind?: string; role?: string; state?: string } }
 export type HistoryPage = {
   epoch: string
   fence?: number
@@ -155,7 +155,7 @@ export async function userItemIds(page: Page, sessionId: string, limit = 200): P
     ids.unshift(
       ...read.page.items
         .filter((item) => item.body?.kind === 'message' && item.body.role === 'user')
-        .map((item) => item.id)
+        .map((item) => item.itemId)
     )
     if (!read.page.hasOlder || ids.length >= limit) {
       return ids.slice(-limit)
@@ -212,17 +212,26 @@ export async function waitForWorking(
   working: boolean,
   timeoutMs = 120_000
 ): Promise<number> {
-  const handle = await page.waitForFunction(
-    (working) => {
-      const root = [...document.querySelectorAll('[data-native-chat-root="true"]')].find(
-        (candidate) => candidate.checkVisibility()
+  const handle = await page
+    .waitForFunction(
+      (working) => {
+        const root = [...document.querySelectorAll('[data-native-chat-root="true"]')].find(
+          (candidate) => candidate.checkVisibility()
+        )
+        return root?.getAttribute('data-native-chat-working') === String(working)
+          ? performance.now()
+          : false
+      },
+      working,
+      { timeout: timeoutMs, polling: 16 }
+    )
+    .catch(async (error: unknown) => {
+      throw new Error(
+        `working=${working} not shown in ${timeoutMs} ms (${await visibleChatState(page)})`,
+        {
+          cause: error
+        }
       )
-      return root?.getAttribute('data-native-chat-working') === String(working)
-        ? performance.now()
-        : false
-    },
-    working,
-    { timeout: timeoutMs, polling: 16 }
-  )
+    })
   return (await handle.jsonValue()) as number
 }