From 8ac7ce73b34a3adbb6b54f568b74b626479fc89d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E4=B8=8D=E5=81=9A=E4=BA=86=E7=9D=A1=E5=A4=A7=E8=A7=89?=
 <64798754+stakeswky@users.noreply.github.com>
Date: Fri, 6 Mar 2026 03:58:21 +0800
Subject: [PATCH] fix: avoid false global rate-limit classification from
 generic cooldown text (#32972)

Merged via squash.

Prepared head SHA: 813c16f5afce415da130a917d9ce9f968912b477
Co-authored-by: stakeswky <64798754+stakeswky@users.noreply.github.com>
Co-authored-by: altaywtf <9790196+altaywtf@users.noreply.github.com>
Reviewed-by: @altaywtf
---
 CHANGELOG.md                                             | 2 ++
 .../pi-embedded-helpers.isbillingerrormessage.test.ts    | 4 +---
 src/agents/pi-embedded-helpers/failover-matches.ts       | 1 -
 ...n-embedded-pi-agent.auth-profile-rotation.e2e.test.ts | 9 +++++++++
 4 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index ab04d677757..1063cd2aea9 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -138,6 +138,8 @@ Docs: https://docs.openclaw.ai
 
 - Mattermost/plugin SDK import policy: replace remaining monolithic `openclaw/plugin-sdk` imports in Mattermost mention-gating paths/tests with scoped subpaths (`openclaw/plugin-sdk/compat` and `openclaw/plugin-sdk/mattermost`) so `pnpm check` passes `lint:plugins:no-monolithic-plugin-sdk-entry-imports` on baseline. (#36480) Thanks @Takhoffman.
 
+- Agents/failover cooldown classification: stop treating generic `cooling down` text as provider `rate_limit` so healthy models no longer show false global cooldown/rate-limit warnings while explicit `model_cooldown` markers still trigger failover. (#32972) thanks @stakeswky.
+
 ## 2026.3.2
 
 ### Changes
diff --git a/src/agents/pi-embedded-helpers.isbillingerrormessage.test.ts b/src/agents/pi-embedded-helpers.isbillingerrormessage.test.ts
index 8d9c678035a..599440ca0b2 100644
--- a/src/agents/pi-embedded-helpers.isbillingerrormessage.test.ts
+++ b/src/agents/pi-embedded-helpers.isbillingerrormessage.test.ts
@@ -498,9 +498,7 @@ describe("classifyFailoverReason", () => {
     expect(
       classifyFailoverReason("model_cooldown: All credentials for model gpt-5 are cooling down"),
     ).toBe("rate_limit");
-    expect(classifyFailoverReason("all credentials for model x are cooling down")).toBe(
-      "rate_limit",
-    );
+    expect(classifyFailoverReason("all credentials for model x are cooling down")).toBeNull();
     expect(
       classifyFailoverReason(
         '{"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}',
diff --git a/src/agents/pi-embedded-helpers/failover-matches.ts b/src/agents/pi-embedded-helpers/failover-matches.ts
index 451852282c6..ecf7be953d9 100644
--- a/src/agents/pi-embedded-helpers/failover-matches.ts
+++ b/src/agents/pi-embedded-helpers/failover-matches.ts
@@ -4,7 +4,6 @@ const ERROR_PATTERNS = {
   rateLimit: [
     /rate[_ ]limit|too many requests|429/,
     "model_cooldown",
-    "cooling down",
     "exceeded your current quota",
     "resource has been exhausted",
     "quota exceeded",
diff --git a/src/agents/pi-embedded-runner.run-embedded-pi-agent.auth-profile-rotation.e2e.test.ts b/src/agents/pi-embedded-runner.run-embedded-pi-agent.auth-profile-rotation.e2e.test.ts
index cf56036c3ea..cfefc20cc67 100644
--- a/src/agents/pi-embedded-runner.run-embedded-pi-agent.auth-profile-rotation.e2e.test.ts
+++ b/src/agents/pi-embedded-runner.run-embedded-pi-agent.auth-profile-rotation.e2e.test.ts
@@ -639,6 +639,15 @@ describe("runEmbeddedPiAgent auth profile rotation", () => {
     expect(typeof usageStats["openai:p2"]?.lastUsed).toBe("number");
   });
 
+  it("rotates for overloaded prompt failures across auto-pinned profiles", async () => {
+    const { usageStats } = await runAutoPinnedRotationCase({
+      errorMessage: '{"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}}',
+      sessionKey: "agent:test:overloaded-rotation",
+      runId: "run:overloaded-rotation",
+    });
+    expect(typeof usageStats["openai:p2"]?.lastUsed).toBe("number");
+  });
+
   it("rotates on timeout without cooling down the timed-out profile", async () => {
     const { usageStats } = await runAutoPinnedRotationCase({
       errorMessage: "request ended without sending any chunks",