diff --git a/src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts b/src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts
index 5346f1a830b..0b4a1c14641 100644
--- a/src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts
+++ b/src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts
@@ -147,6 +147,14 @@ describe("formatAssistantErrorText", () => {
     );
   });
 
+  it("strips leading HTTP status code prefix from non-JSON rate limit messages", () => {
+    const msg = makeAssistantError("429 Your quota has been exhausted, try again in 24 hours");
+    const result = formatAssistantErrorText(msg);
+    expect(result).toContain("try again in 24 hours");
+    expect(result).not.toMatch(/^⚠️ 429\b/);
+    expect(result).toBe("⚠️ Your quota has been exhausted, try again in 24 hours");
+  });
+
   it("returns a friendly message for empty stream chunk errors", () => {
     const msg = makeAssistantError("request ended without sending any chunks");
     expect(formatAssistantErrorText(msg)).toBe("LLM request timed out.");
diff --git a/src/agents/pi-embedded-helpers/errors.ts b/src/agents/pi-embedded-helpers/errors.ts
index b7df985e379..0132fa33998 100644
--- a/src/agents/pi-embedded-helpers/errors.ts
+++ b/src/agents/pi-embedded-helpers/errors.ts
@@ -68,7 +68,9 @@ const RATE_LIMIT_SPECIFIC_HINT_RE =
 function extractProviderRateLimitMessage(raw: string): string | undefined {
   // Try to pull a human-readable message out of a JSON error payload first.
   const info = parseApiErrorInfo(raw);
-  const candidate = info?.message ?? raw;
+  // When the raw string is not a JSON payload, strip any leading HTTP status
+  // code (e.g. "429 ") so the surfaced message stays clean.
+  const candidate = info?.message ?? (extractLeadingHttpStatus(raw.trim())?.rest || raw);
 
   if (!candidate || !RATE_LIMIT_SPECIFIC_HINT_RE.test(candidate)) {
     return undefined;