fix(agent): classify message-only 'overloaded' as server overload

Salvage of #14261 by @ms-alan — rebased onto current main, scoped to the
overloaded-classification fix, with a regression test that fails without it.

agent/error_classifier.py

@@ -1214,6 +1214,20 @@ def _classify_by_message(
             should_fallback=True,
         )
 
+    # Overloaded / server-busy patterns — must come BEFORE the rate_limit and
+    # billing checks so that a message-only "overloaded" (no 503/529 status,
+    # e.g. some Anthropic-compatible proxies) classifies as a transient
+    # overload (backoff + retry) instead of falling through to `unknown` or
+    # incorrectly triggering credential rotation.
+    if any(p in error_msg for p in (
+        "overloaded", "temporarily overloaded",
+        "service is temporarily overloaded",
+    )):
+        return result_fn(
+            FailoverReason.overloaded,
+            retryable=True,
+        )
+
     # Billing patterns
     if any(p in error_msg for p in _BILLING_PATTERNS):
         return result_fn(

tests/agent/test_error_classifier.py

@@ -347,6 +347,18 @@ class TestClassifyApiError:
         result = classify_api_error(e)
         assert result.reason == FailoverReason.overloaded
 
+    def test_message_only_overloaded_without_status_is_overloaded(self):
+        """Some Anthropic-compatible proxies surface 'overloaded' in the
+        message with no 503/529 status_code. It must classify as overloaded
+        (transient backoff+retry), not unknown / credential rotation. (#14261)"""
+        e = MockAPIError(
+            "Anthropic API error: Overloaded - the service is temporarily overloaded"
+        )  # no status_code
+        result = classify_api_error(e, provider="anthropic")
+        assert result.reason == FailoverReason.overloaded
+        assert result.retryable is True
+        assert result.should_rotate_credential is False
+
     # ── 5xx that are actually request-validation errors ──
     # Some OpenAI-compatible gateways (e.g. codex.nekos.me) return
     # request-validation failures with a 5xx status. These are