From de82d17de2a7964bfa05988698f019ffe7801f1e Mon Sep 17 00:00:00 2001
From: Vincent Koc <vincentkoc@ieee.org>
Date: Sun, 3 May 2026 16:50:32 -0700
Subject: [PATCH] fix(qa): require cache probe marker

---
 CHANGELOG.md                                    |  1 +
 src/agents/live-cache-regression-runner.test.ts |  7 +++++++
 src/agents/live-cache-regression-runner.ts      | 13 +++++++++----
 3 files changed, 17 insertions(+), 4 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index a06e139a48a..7e3a7ebba88 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -35,6 +35,7 @@ Docs: https://docs.openclaw.ai
 
 - Channels/WhatsApp: allow `@whiskeysockets/libsignal-node` in `onlyBuiltDependencies` so pnpm v9+ `blockExoticSubdeps` no longer rejects the baileys git-tarball subdep and silences all inbound agent replies. Fixes #76539. Thanks @ottodeng and @vincentkoc.
 - Gateway/systemd: preserve operator-added secrets in the Gateway env file across re-stage while clearing OpenClaw-managed keys (such as `OPENCLAW_GATEWAY_TOKEN`) so a fresh staging value is never shadowed by a stale env-file copy; operator secrets are also retained when the state-dir `.env` is empty. Fixes #76860. Thanks @hclsys.
+- QA/cache: require the full `CACHE-OK <suffix>` marker before live cache probes stop retrying, so suffix-only prose cannot hide a broken probe response. Thanks @vincentkoc.
 - Slack/Matrix: avoid creating blank progress-draft messages when `streaming.progress.label=false` and progress tool lines are disabled. Thanks @vincentkoc.
 - QA/Matrix: keep the mock OpenAI tool-progress provider aligned with exact-marker Matrix prompts so the hardened live preview scenario still forces a deterministic read before final delivery. Thanks @vincentkoc.
 - OpenAI/Google Meet: wait for realtime voice `session.updated` before treating the bridge as connected, so Meet joins do not return with audio queued behind an unconfigured realtime session. Thanks @vincentkoc.
diff --git a/src/agents/live-cache-regression-runner.test.ts b/src/agents/live-cache-regression-runner.test.ts
index ecd30005732..97bc267836f 100644
--- a/src/agents/live-cache-regression-runner.test.ts
+++ b/src/agents/live-cache-regression-runner.test.ts
@@ -99,6 +99,13 @@ describe("live cache regression runner", () => {
         text: "",
       }),
     ).toBe(false);
+    expect(
+      __testing.shouldRetryCacheProbeText({
+        attempt: 1,
+        suffix: "openai-stable-hit-a",
+        text: "I saw openai-stable-hit-a.",
+      }),
+    ).toBe(true);
     expect(
       __testing.shouldRetryCacheProbeText({
         attempt: 1,
diff --git a/src/agents/live-cache-regression-runner.ts b/src/agents/live-cache-regression-runner.ts
index f47fbebc1fe..c0c54d046d9 100644
--- a/src/agents/live-cache-regression-runner.ts
+++ b/src/agents/live-cache-regression-runner.ts
@@ -136,7 +136,11 @@ function shouldRetryCacheProbeText(params: {
 }): boolean {
   const responseTextLower = normalizeLowercaseStringOrEmpty(params.text);
   const suffixLower = normalizeLowercaseStringOrEmpty(params.suffix);
-  return !responseTextLower.includes(suffixLower) && params.attempt <= LIVE_CACHE_RESPONSE_RETRIES;
+  const markerLower = `cache-ok ${suffixLower}`;
+  return (
+    (!responseTextLower.includes(markerLower) || !responseTextLower.includes(suffixLower)) &&
+    params.attempt <= LIVE_CACHE_RESPONSE_RETRIES
+  );
 }
 
 async function runToolOnlyTurn(params: {
@@ -244,9 +248,10 @@ async function completeCacheProbe(params: {
     }
     const responseTextLower = normalizeLowercaseStringOrEmpty(text);
     const suffixLower = normalizeLowercaseStringOrEmpty(params.suffix);
+    const markerLower = `cache-ok ${suffixLower}`;
     assert(
-      responseTextLower.includes(suffixLower),
-      `expected response to contain ${params.suffix}, got ${JSON.stringify(text)}`,
+      responseTextLower.includes(markerLower),
+      `expected response to contain CACHE-OK ${params.suffix}, got ${JSON.stringify(text)}`,
     );
     const usage = normalizeCacheUsage(response.usage);
     return {
@@ -256,7 +261,7 @@ async function completeCacheProbe(params: {
       hitRate: computeCacheHitRate(usage),
     };
   }
-  throw new Error(`expected response to contain ${params.suffix}`);
+  throw new Error(`expected response to contain CACHE-OK ${params.suffix}`);
 }
 
 async function runRepeatedLane(params: {