fix(anthropic): OAuth token endpoint UA must not be claude-code/ (login 429, #48534)

agent/anthropic_adapter.py

@@ -1045,7 +1045,7 @@ def refresh_anthropic_oauth_pure(refresh_token: str, *, use_json: bool = False)
             data=data,
             headers={
                 "Content-Type": content_type,
-                "User-Agent": f"claude-code/{_get_claude_code_version()} (external, cli)",
+                "User-Agent": _OAUTH_TOKEN_USER_AGENT,
             },
             method="POST",
         )
@@ -1378,6 +1378,16 @@ def run_oauth_setup_token() -> Optional[str]:
     "https://console.anthropic.com/v1/oauth/token",
 ]
 _OAUTH_TOKEN_URL = _OAUTH_TOKEN_URLS[0]
+# User-Agent sent on the OAuth *token endpoint* (login exchange + refresh).
+# Anthropic rate-limits (HTTP 429) any token-endpoint request whose UA starts
+# with ``claude-code/`` — verified empirically against platform.claude.com:
+# ``claude-code/2.1.200`` and ``Mozilla/5.0`` -> 429; ``axios/*``, ``node``,
+# and SDK-style UAs -> 400 (reached code validation). The real Claude Code CLI
+# exchanges the auth code with a bare axios client (``axios/<ver>``), NOT its
+# ``claude-code/`` inference UA. We mirror that here. NOTE: the *inference* path
+# (build_anthropic_kwargs) still uses the ``claude-code/`` UA + ``x-app: cli`` —
+# that fingerprint is required there and is NOT throttled on the messages API.
+_OAUTH_TOKEN_USER_AGENT = "axios/1.7.9"
 _OAUTH_REDIRECT_URI = "https://console.anthropic.com/oauth/code/callback"
 _OAUTH_SCOPES = "org:create_api_key user:profile user:inference"
 _HERMES_OAUTH_FILE = get_hermes_home() / ".anthropic_oauth.json"
@@ -1478,8 +1488,9 @@ def run_hermes_oauth_login_pure() -> Optional[Dict[str, Any]]:
         # Anthropic migrated the OAuth token endpoint to platform.claude.com;
         # console.anthropic.com now 404s. Try the new host first, then fall
         # back to console for older deployments (mirrors the refresh path).
-        # Use the claude-code/ UA prefix: Anthropic blocks claude-cli/ on the
-        # OAuth token endpoint (returns 404 for all versions).
+        # UA is _OAUTH_TOKEN_USER_AGENT (a non-claude-code UA) — see the
+        # constant's definition for why the token endpoint must not send
+        # claude-code/ (429 UA-prefix block).
         result = None
         last_error = None
         for endpoint in _OAUTH_TOKEN_URLS:
@@ -1488,7 +1499,7 @@ def run_hermes_oauth_login_pure() -> Optional[Dict[str, Any]]:
                 data=exchange_data,
                 headers={
                     "Content-Type": "application/json",
-                    "User-Agent": f"claude-code/{_get_claude_code_version()} (external, cli)",
+                    "User-Agent": _OAUTH_TOKEN_USER_AGENT,
                 },
                 method="POST",
             )

tests/agent/test_anthropic_adapter.py

@@ -507,12 +507,22 @@ def test_keeps_static_anthropic_token_when_only_non_refreshable_claude_key_exist
 
 
 class TestRefreshOauthToken:
-    def test_returns_none_without_refresh_token(self):
+    def test_returns_none_without_refresh_token(self, tmp_path, monkeypatch):
+        monkeypatch.setattr("agent.anthropic_adapter.Path.home", lambda: tmp_path)
+        # Neutralize live Claude Code sources (macOS Keychain + ~/.claude file)
+        # so the adopt-already-refreshed branch can't short-circuit with a real
+        # credential on a dev/CI machine that happens to have Claude Code creds.
+        monkeypatch.setattr(
+            "agent.anthropic_adapter.read_claude_code_credentials", lambda: None
+        )
         creds = {"accessToken": "expired", "refreshToken": "", "expiresAt": 0}
         assert _refresh_oauth_token(creds) is None
 
     def test_successful_refresh(self, tmp_path, monkeypatch):
         monkeypatch.setattr("agent.anthropic_adapter.Path.home", lambda: tmp_path)
+        monkeypatch.setattr(
+            "agent.anthropic_adapter.read_claude_code_credentials", lambda: None
+        )
 
         creds = {
             "accessToken": "old-token",
@@ -544,7 +554,11 @@ def test_successful_refresh(self, tmp_path, monkeypatch):
         assert written["claudeAiOauth"]["accessToken"] == "new-token-abc"
         assert written["claudeAiOauth"]["refreshToken"] == "new-refresh-456"
 
-    def test_failed_refresh_returns_none(self):
+    def test_failed_refresh_returns_none(self, tmp_path, monkeypatch):
+        monkeypatch.setattr("agent.anthropic_adapter.Path.home", lambda: tmp_path)
+        monkeypatch.setattr(
+            "agent.anthropic_adapter.read_claude_code_credentials", lambda: None
+        )
         creds = {
             "accessToken": "old",
             "refreshToken": "refresh-123",

tests/agent/test_anthropic_oauth_ua_prefix.py

@@ -1,7 +1,16 @@
 """Regression tests for the OAuth User-Agent header in anthropic_adapter.py.
 
-Anthropic now 404s the OAuth token endpoint for any ``claude-cli/`` UA prefix
-(issue #48534). The adapter must use ``claude-code/`` instead.
+Two DIFFERENT Anthropic endpoints impose OPPOSITE User-Agent requirements:
+
+- Inference (``/v1/messages`` via build_anthropic_client): requires the
+  ``claude-code/`` UA + ``x-app: cli`` fingerprint, or requests get
+  intermittent 500s. (issue #48534: ``claude-cli/`` is 404'd here.)
+- OAuth token endpoint (``/v1/oauth/token`` login exchange + refresh):
+  Anthropic now RATE-LIMITS (HTTP 429) any UA whose prefix is ``claude-code/``
+  (or ``Mozilla/``). Verified empirically against platform.claude.com:
+  ``claude-code/2.1.200`` -> 429; ``axios/*`` / ``node`` -> 400 (reached code
+  validation). The token endpoint must therefore use a non-``claude-code/`` UA
+  (we send ``axios/*``, matching the real Claude Code CLI's exchange client).
 """
 
 from __future__ import annotations
@@ -13,10 +22,10 @@
 
 
 class TestOAuthUserAgentPrefix:
-    """All OAuth-related HTTP requests must use ``claude-code/`` UA, not ``claude-cli/``."""
+    """Inference uses ``claude-code/``; the OAuth token endpoint must NOT."""
 
     def test_build_anthropic_client_oauth_ua(self):
-        """build_anthropic_client with OAuth token must use claude-code UA."""
+        """build_anthropic_client (INFERENCE) with OAuth token must use claude-code UA."""
         from agent.anthropic_adapter import build_anthropic_client
 
         mock_sdk = MagicMock()
@@ -47,33 +56,40 @@ def test_no_claude_cli_in_source(self):
                     f"Line {i}: claude-cli/ still used in User-Agent header: {stripped}"
                 )
 
-    def test_token_exchange_ua_prefix(self):
-        """run_hermes_oauth_login_pure must not send claude-cli/ UA."""
+    def test_token_exchange_ua_not_throttled(self):
+        """run_hermes_oauth_login_pure must NOT send a throttled token-endpoint UA.
+
+        Anthropic 429s both ``claude-cli/`` and ``claude-code/`` UAs at the
+        token endpoint. The login exchange must use the shared
+        ``_OAUTH_TOKEN_USER_AGENT`` constant (a non-claude-code UA).
+        """
         import inspect
         import agent.anthropic_adapter as mod
 
-        # Get the source of the exchange function
         try:
             source = inspect.getsource(mod.run_hermes_oauth_login_pure)
         except AttributeError:
             pytest.skip("run_hermes_oauth_login_pure not found")
 
-        # Only fail on claude-cli/ in an actual User-Agent header line — a
-        # comment that references the old behavior (e.g. "Anthropic blocks
-        # claude-cli/ on the OAuth endpoint") is allowed. Mirrors the
-        # header-scoped check in test_no_claude_cli_in_source.
         for i, line in enumerate(source.split("\n"), 1):
             stripped = line.strip()
-            if "claude-cli/" in stripped and ("User-Agent" in stripped or "user-agent" in stripped):
+            if ("User-Agent" in stripped or "user-agent" in stripped) and (
+                "claude-cli/" in stripped or "claude-code/" in stripped
+            ):
                 pytest.fail(
-                    f"Line {i}: run_hermes_oauth_login_pure still uses claude-cli/ UA header: {stripped}"
+                    f"Line {i}: throttled UA in token-exchange header: {stripped}"
                 )
-        assert "claude-code/" in source, (
-            "run_hermes_oauth_login_pure should use claude-code/ UA"
+        assert "_OAUTH_TOKEN_USER_AGENT" in source, (
+            "run_hermes_oauth_login_pure should send the shared "
+            "_OAUTH_TOKEN_USER_AGENT (non-claude-code) on the token endpoint"
+        )
+        assert not mod._OAUTH_TOKEN_USER_AGENT.startswith(("claude-code/", "claude-cli/")), (
+            f"_OAUTH_TOKEN_USER_AGENT must not be a throttled prefix: "
+            f"{mod._OAUTH_TOKEN_USER_AGENT!r}"
         )
 
-    def test_token_refresh_ua_prefix(self):
-        """refresh_anthropic_oauth_pure must not send claude-cli/ UA."""
+    def test_token_refresh_ua_not_throttled(self):
+        """refresh_anthropic_oauth_pure must NOT send a throttled token-endpoint UA."""
         import inspect
         import agent.anthropic_adapter as mod
 
@@ -82,13 +98,15 @@ def test_token_refresh_ua_prefix(self):
             pytest.skip("refresh_anthropic_oauth_pure not found")
         source = inspect.getsource(func)
 
-        # Header-scoped check (comments referencing claude-cli/ are allowed).
         for i, line in enumerate(source.split("\n"), 1):
             stripped = line.strip()
-            if "claude-cli/" in stripped and ("User-Agent" in stripped or "user-agent" in stripped):
+            if ("User-Agent" in stripped or "user-agent" in stripped) and (
+                "claude-cli/" in stripped or "claude-code/" in stripped
+            ):
                 pytest.fail(
-                    f"Line {i}: refresh_anthropic_oauth_pure still uses claude-cli/ UA header: {stripped}"
+                    f"Line {i}: throttled UA in refresh header: {stripped}"
                 )
-        assert "claude-code/" in source, (
-            "refresh_anthropic_oauth_pure should use claude-code/ UA"
+        assert "_OAUTH_TOKEN_USER_AGENT" in source, (
+            "refresh_anthropic_oauth_pure should send the shared "
+            "_OAUTH_TOKEN_USER_AGENT (non-claude-code) on the token endpoint"
         )