test(integration): name Claude Code wire tests by behavior and move provider-agnostic ones to routing and streaming

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

tests/integration/README.md

@@ -30,6 +30,8 @@ Streaming checks send real HTTP transfer chunks, including one-byte partitions,
 
 The `messages_endpoint/` directory holds `/v1/messages` endpoint contracts: native-provider backends under `providers/` (`anthropic`, `bedrock`, `gemini`) and the translation bridges (`responses_bridge`, `chat_bridge`) at the top level. It runs in the providers shard; `run.py` selects test files recursively under each scheduled directory
 
+A provider folder holds only what depends on that provider's wire format: request and response fidelity, multi-turn history, request parameters, content types, and provider-specific pricing. Behavior every provider shares, such as fallback or billing after a client disconnect, lives in the feature directory it exercises (`routing/`, `streaming/`, `spend/`). `_support/claude_code.py` holds a captured Claude Code request and stream builders that any directory can use as a realistic agent payload
+
 The sdk shard exercises the SDK's own HTTP clients against local protocol peers with no gateway in the path, so a case here fails only when the client library or its wire behavior changes. The HTTP/2 case runs a hypercorn TLS peer offering h2 and http/1.1 over ALPN, drives the sync and async httpx handlers at it with `LITELLM_HTTP2` off and on, and asserts the version both the client and the peer observed on the wire. Put a test here only when it needs no proxy, database or Redis; a case that reaches the gateway belongs in one of the other shards
 
 The extensions shard uses the built-in generic callback and guardrail transports. It checks callback correlation and credential exclusion, guardrail rewriting and denial, retained OpenAI consumers and A2A wire versions

tests/integration/_support/claude_code.py

@@ -1,4 +1,4 @@
-"""Shared Claude Code-shaped request builders and upstream stream fixtures for the /v1/messages contracts."""
+"""Shared Claude Code-shaped request builders and upstream stream fixtures for integration contracts."""
 
 import json
 from collections.abc import Mapping

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_adaptive_thinking_effort_wire.py

@@ -5,7 +5,7 @@ import pytest
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 
@@ -17,7 +17,7 @@ def _diff(expected: dict[str, JsonValue], body: dict[str, JsonValue]) -> dict[st
     }
 
 
-def test_claude_code_adaptive_thinking_and_effort_reach_anthropic_intact(gateway: Gateway) -> None:
+def test_adaptive_thinking_and_effort_reach_anthropic_intact(gateway: Gateway) -> None:
     identity: Final = f"msg_fable_{uuid.uuid4().hex}"
     request_body: Final = cc.frontier_request(f"cache-bust-{uuid.uuid4().hex}", "high", 64000)
     cli_beta: Final = frozenset(cc.FRONTIER_CLI_BETA.split(","))
@@ -63,7 +63,7 @@ def test_claude_code_adaptive_thinking_and_effort_reach_anthropic_intact(gateway
         assert rows[0]["prompt_tokens"] == 12 and rows[0]["completion_tokens"] == 4
 
 
-def test_claude_code_xhigh_effort_reaches_anthropic_and_charges_by_usage(gateway: Gateway) -> None:
+def test_xhigh_effort_reaches_anthropic_and_charges_by_usage(gateway: Gateway) -> None:
     identity: Final = f"msg_opus_{uuid.uuid4().hex}"
     request_body: Final = cc.frontier_request(f"cache-bust-{uuid.uuid4().hex}", "xhigh", 128000)
     cli_beta: Final = frozenset(cc.FRONTIER_CLI_BETA.split(","))

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_compaction_wire.py

@@ -3,7 +3,7 @@ from typing import Final
 
 from integration._support.client import Gateway
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 _CONTEXT_MANAGEMENT: Final = {

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_document_input_wire.py

@@ -4,7 +4,7 @@ from typing import Final
 
 from integration._support.client import Gateway
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 _PDF_BYTES: Final = (

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_image_input_wire.py

@@ -3,7 +3,7 @@ from typing import Final
 
 from integration._support.client import Gateway
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 _PNG_B64: Final = "iVBORw0KGgoAAAANSUhEUgAAAAQAAAAECAIAAAAmkwkpAAAAEElEQVR4nGP4z8AARwzEcQCukw/x0F8jngAAAABJRU5ErkJggg=="

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_interleaved_thinking_history_wire.py

@@ -3,7 +3,7 @@ from typing import Final
 
 from integration._support.client import Gateway
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 _TURN1_TOOL: Final = ("toolu_a", "Read", {"file_path": "/tmp/cc_probe/a.txt"})

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_long_context_beta_wire.py

@@ -5,7 +5,7 @@ import pytest
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 
 _BETA_1M: Final = f"{cc.FRONTIER_CLI_BETA.replace(',effort-2025-11-24', ',context-1m-2025-08-07,effort-2025-11-24')}"
 

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_model_switch_history_wire.py

@@ -4,11 +4,11 @@ from typing import Final
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 
-def test_claude_code_mid_loop_model_switch_replays_history_byte_identical(gateway: Gateway) -> None:
+def test_mid_loop_model_switch_replays_history_byte_identical(gateway: Gateway) -> None:
     identity1: Final = f"msg_sw1_{uuid.uuid4().hex}"
     identity2: Final = f"msg_sw2_{uuid.uuid4().hex}"
     turn1: Final = cc.frontier_request(

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_prompt_cache_billing_wire.py

@@ -5,7 +5,7 @@ import pytest
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 _USAGE: Final = {
@@ -16,7 +16,7 @@ _USAGE: Final = {
 }
 
 
-def test_claude_code_cached_turn_charges_cache_read_and_creation_rates(gateway: Gateway) -> None:
+def test_cached_turn_charges_cache_read_and_creation_rates(gateway: Gateway) -> None:
     identity: Final = f"msg_pc_{uuid.uuid4().hex}"
     turn1: Final = cc.frontier_request(
         f"cache-bust-{uuid.uuid4().hex}",

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_request_fidelity_wire.py

@@ -4,12 +4,12 @@ from typing import Final
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 
 _MODEL: Final = cc.SONNET
 
 
-def test_claude_code_streaming_request_reaches_anthropic_intact_and_streams_back(gateway: Gateway) -> None:
+def test_streaming_request_reaches_anthropic_intact_and_streams_back(gateway: Gateway) -> None:
     identity: Final = f"msg_cc_{uuid.uuid4().hex}"
     request_body: Final = cc.claude_code_request(f"cache-bust-{uuid.uuid4().hex}")
     cli_beta: Final = frozenset(cc.CLI_BETA.split(","))

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_tool_loop_wire.py

@@ -4,7 +4,7 @@ from typing import Final
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 from pydantic import JsonValue
 
 _THINKING: Final = "need to read the file"
@@ -20,7 +20,7 @@ def _diff(expected: dict[str, JsonValue], body: dict[str, JsonValue]) -> dict[st
     }
 
 
-def test_claude_code_tool_loop_round_trips_thinking_tool_use_and_tool_result(gateway: Gateway) -> None:
+def test_tool_loop_round_trips_thinking_tool_use_and_tool_result(gateway: Gateway) -> None:
     identity1: Final = f"msg_tl1_{uuid.uuid4().hex}"
     identity2: Final = f"msg_tl2_{uuid.uuid4().hex}"
     turn1: Final = cc.frontier_request(
@@ -114,7 +114,7 @@ def test_claude_code_tool_loop_round_trips_thinking_tool_use_and_tool_result(gat
         assert rows[0]["prompt_tokens"] == 30
 
 
-def test_claude_code_parallel_tool_results_reach_anthropic_in_client_order(gateway: Gateway) -> None:
+def test_parallel_tool_results_reach_anthropic_in_client_order(gateway: Gateway) -> None:
     turn1: Final = cc.frontier_request(
         f"cache-bust-{uuid.uuid4().hex}",
         "high",

tests/integration/messages_endpoint/providers/anthropic/test_anthropic_web_search_citations_wire.py

@@ -4,7 +4,7 @@ from typing import Final
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 
 WEB_SEARCH_TOOL: Final = {"type": "web_search_20250305", "name": "web_search", "max_uses": 8}
 
@@ -103,7 +103,7 @@ def _web_search_stream(identity: str) -> tuple[bytes, ...]:
     )
 
 
-def test_claude_code_web_search_tool_passthrough_and_cited_response(gateway: Gateway) -> None:
+def test_web_search_tool_passthrough_and_cited_response(gateway: Gateway) -> None:
     identity: Final = f"msg_ws_{uuid.uuid4().hex}"
     base: Final = cc.frontier_request(
         f"cache-bust-{uuid.uuid4().hex}",

tests/integration/routing/test_overloaded_deployment_fallback.py

@@ -9,7 +9,7 @@ from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.process import owned_proxy
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 
 
 def _error_529() -> Reply:
@@ -19,7 +19,7 @@ def _error_529() -> Reply:
     )
 
 
-def test_anthropic_overloaded_primary_falls_back_to_second_deployment(gateway: Gateway, tmp_path: Path) -> None:
+def test_overloaded_primary_falls_back_to_second_deployment(gateway: Gateway, tmp_path: Path) -> None:
     identity: Final = f"msg_fb_{uuid.uuid4().hex}"
     request_body: Final = cc.claude_code_request(f"cache-bust-{uuid.uuid4().hex}")
 

tests/integration/streaming/test_client_disconnect_still_bills.py

@@ -4,7 +4,7 @@ from typing import Final
 from integration._support.client import Gateway, eventually
 from integration._support.database import read_rows
 from integration._support.wire import Reply, Request, wire_server
-from integration.messages_endpoint import _claude_code as cc
+from integration._support import claude_code as cc
 
 
 def test_client_disconnect_mid_stream_still_bills_the_message(gateway: Gateway) -> None: