Merge 2bfcbaae04 into b40bd7bf03

2026-10-26 06:03:19 +00:00 · 2026-10-19 11:32:15 -07:00 · 2026-10-19 11:32:15 -07:00 · f6337ce0d5
commit f6337ce0d5
parent b40bd7bf03 2bfcbaae04
7 changed files with 48 additions and 20 deletions
--- a/litellm/constants.py
+++ b/litellm/constants.py
@ -186,6 +186,7 @@ MCP_PER_USER_TOKEN_EXPIRY_BUFFER_SECONDS: Final = int(os.getenv("MCP_PER_USER_TO

 # MCP timeout defaults (seconds). Override via env vars for slow/custom MCP servers.
 MCP_CLIENT_TIMEOUT: Final = float(os.getenv("LITELLM_MCP_CLIENT_TIMEOUT", "60.0"))
+MCP_MAX_SSE_EVENT_SIZE_BYTES: Final = get_env_int("LITELLM_MCP_MAX_SSE_EVENT_SIZE_BYTES", 64 * 1024 * 1024)
 MCP_TOOL_LISTING_TIMEOUT: Final = float(os.getenv("LITELLM_MCP_TOOL_LISTING_TIMEOUT", "30.0"))
 MCP_METADATA_TIMEOUT: Final = float(os.getenv("LITELLM_MCP_METADATA_TIMEOUT", "10.0"))
 MCP_HEALTH_CHECK_TIMEOUT: Final = float(os.getenv("LITELLM_MCP_HEALTH_CHECK_TIMEOUT", "10.0"))
--- a/litellm/experimental_mcp_client/client.py
+++ b/litellm/experimental_mcp_client/client.py
@ -54,6 +54,7 @@ from pydantic import AnyUrl
 from litellm._logging import verbose_logger
 from litellm.constants import (
    MCP_CLIENT_TIMEOUT,
+    MCP_MAX_SSE_EVENT_SIZE_BYTES,
    MCP_NPM_CACHE_DIR,
    MCP_TOOL_LISTING_MAX_PAGES,
    MCP_TOOL_LISTING_TIMEOUT,
@ -475,6 +476,7 @@ class MCPClient:
        transport_ctx: Final = streamable_http_client(
            url=self.server_url,
            http_client=http_client,
+            max_sse_event_size=MCP_MAX_SSE_EVENT_SIZE_BYTES,
        )
        return transport_ctx, http_client

--- a/pyproject.toml
+++ b/pyproject.toml
@ -72,8 +72,8 @@ proxy = [
    "boto3>=1.43.1,<2.0",
    "azure-identity>=1.25.2,<2.0",
    "azure-storage-blob>=12.28.0,<13.0",
-    "mcp>=2.2.0,<3",
-    "httpx2>=2.5.0,<3",
+    "mcp>=2.3.0,<3",
+    "httpx2>=2.10.0,<3",
    "pydantic>=2.12.0,<3",
    "litellm-proxy-extras==0.4.102.post1",
    "litellm-enterprise==0.1.71",
@ -121,7 +121,7 @@ utils = [
    "numpydoc>=1.8.0,<2.0",
 ]
 caching = ["diskcache>=5.6.3,<6.0"]
-mcp = ["mcp>=2.2.0,<3", "httpx2>=2.5.0,<3", "pydantic>=2.12.0,<3"]
+mcp = ["mcp>=2.3.0,<3", "httpx2>=2.10.0,<3", "pydantic>=2.12.0,<3"]
 # Driver for the MongoDB Atlas vector store; Atlas Vector Search has no HTTP query API.
 # The floor is 4.9 because that is the release AsyncMongoClient landed in.
 # SAML SSO for the admin UI. python3-saml pulls in xmlsec/lxml, whose wheels
@ -243,7 +243,7 @@ e2e-dev = [
    "locust==2.45.0",
    "anthropic==0.84.0",
    "psutil==7.2.2",
-    "mcp>=2.2.0,<3",
+    "mcp>=2.3.0,<3",
 ]
 proxy-dev = [
    "prisma==0.11.0",
@ -296,7 +296,7 @@ healthcheck = [
 benchmarks = [
    "pytest==9.0.3",
    "pytest-codspeed==4.3.0",
-    "mcp>=2.2.0,<3",
+    "mcp>=2.3.0,<3",
    "a2a-sdk==1.1.0",
 ]

--- a/tests/integration/_support/mcp.py
+++ b/tests/integration/_support/mcp.py
@ -216,6 +216,7 @@ JsonRpc = Mapping[str, object]
 class ScriptedTool:
    name: str
    respond: Callable[[JsonRpc], Reply | JsonRpc]
+    event_stream: bool = False


 def jsonrpc_reply(identity: object, result: JsonRpc) -> Reply:
@ -262,7 +263,10 @@ def scripted_peer(*tools: ScriptedTool) -> Iterator[McpPeer]:
        if tool is None:
            return jsonrpc_error(identity, -32602, "unknown tool")
        produced: Final = tool.respond(body["params"])
-        return produced if isinstance(produced, Reply) else jsonrpc_reply(identity, produced)
+        reply: Final = produced if isinstance(produced, Reply) else jsonrpc_reply(identity, produced)
+        if not tool.event_stream:
+            return reply
+        return Reply(content_type="text/event-stream", body=b"event: message\ndata: " + reply.body + b"\n\n")

    with wire_server(provider) as wire:
        yield McpPeer(wire.url + "/mcp", observed)
--- a/tests/integration/mcp/test_mcp_transports.py
+++ b/tests/integration/mcp/test_mcp_transports.py
@ -10,10 +10,15 @@ from integration._support.mcp import (
    EntryPoint,
    McpCaller,
    PeerKind,
+    ScriptedTool,
+    call_tool,
+    listed_tools,
    mcp_peer,
    official_client_outcomes,
    peer_of,
    register_mcp,
+    scripted_peer,
+    text_result,
    tool_calls,
 )

@ -155,3 +160,19 @@ def test_server_initiated_sampling_and_elicitation_surface_as_errors_not_success
        assert outcome.error is not None, outcome.raw
        assert outcome.text is None or not outcome.text.startswith(("sampled:", "elicited:")), outcome.raw
        assert len(tool_calls(peer.drain())) == 1
+
+
+def test_tool_result_larger_than_one_mib_in_a_single_sse_event_reaches_the_caller(gateway: Gateway) -> None:
+    payload: Final = "x" * (3 * 1024 * 1024)
+    tool: Final = ScriptedTool("dump", lambda _: text_result(payload), event_stream=True)
+
+    with scripted_peer(tool) as peer, gateway.scenario() as scenario:
+        alias: Final = "bigsse" + uuid.uuid4().hex[:8]
+        identity: Final = register_mcp(scenario, peer, alias)
+        key: Final = scenario.key(object_permission={"mcp_servers": [identity]})
+        name: Final = next(full for full in listed_tools(gateway, key, identity) if full.endswith("dump"))
+        response: Final = call_tool(gateway, key, identity, name, {})
+        assert response.status_code == 200, response.text[:500]
+        body: Final = response.json()
+        assert body["isError"] is False, response.text[:500]
+        assert body["content"][0]["text"] == payload
--- a/tests/unit/experimental_mcp_client/test_mcp_client.py
+++ b/tests/unit/experimental_mcp_client/test_mcp_client.py
@ -1236,8 +1236,8 @@ def test_mcp_extra_matches_proxy_extra_and_supports_streamable_http():
    }

    specifier: Final = Requirement(mcp_extra["mcp"]).specifier
-    assert not specifier.contains("1.28.1")
-    assert specifier.contains("2.2.0")
+    assert not specifier.contains("2.2.0")
+    assert specifier.contains("2.3.0")
    with (pyproject_path.parent / "uv.lock").open("rb") as f:
        locked = tomllib.load(f)
    mcp_versions: Final = [package["version"] for package in locked["package"] if package["name"] == "mcp"]
--- a/uv.lock
+++ b/uv.lock
@ -4786,8 +4786,8 @@ requires-dist = [
    { name = "gunicorn", marker = "extra == 'proxy'", specifier = ">=23.0.0,<24.0" },
    { name = "hiredis", marker = "extra == 'proxy'", specifier = ">=3.0.0,<4.0" },
    { name = "httpx", extras = ["http2"], specifier = ">=0.28.0,<1.0" },
-    { name = "httpx2", marker = "extra == 'mcp'", specifier = ">=2.5.0,<3" },
-    { name = "httpx2", marker = "extra == 'proxy'", specifier = ">=2.5.0,<3" },
+    { name = "httpx2", marker = "extra == 'mcp'", specifier = ">=2.10.0,<3" },
+    { name = "httpx2", marker = "extra == 'proxy'", specifier = ">=2.10.0,<3" },
    { name = "huggingface-hub", specifier = ">=0.34.0,<2.0" },
    { name = "importlib-metadata", specifier = ">=8.0.0,<9.0" },
    { name = "inquirerpy", marker = "extra == 'cli'", specifier = ">=0.3.4,<1.0" },
@ -4800,8 +4800,8 @@ requires-dist = [
    { name = "litellm-proxy-extras", marker = "extra == 'proxy'", editable = "litellm-proxy-extras" },
    { name = "llm-sandbox", marker = "extra == 'proxy-runtime'", specifier = ">=0.3.39,<1.0" },
    { name = "mangum", marker = "extra == 'proxy-runtime'", specifier = ">=0.17.0,<1.0" },
-    { name = "mcp", marker = "extra == 'mcp'", specifier = ">=2.2.0,<3" },
-    { name = "mcp", marker = "extra == 'proxy'", specifier = ">=2.2.0,<3" },
+    { name = "mcp", marker = "extra == 'mcp'", specifier = ">=2.3.0,<3" },
+    { name = "mcp", marker = "extra == 'proxy'", specifier = ">=2.3.0,<3" },
    { name = "mlflow", marker = "extra == 'mlflow'", specifier = ">=3.11.1,<4.0" },
    { name = "numpy", marker = "extra == 'stt-nvidia-riva'", specifier = ">=1.26.0" },
    { name = "numpydoc", marker = "extra == 'utils'", specifier = ">=1.8.0,<2.0" },
@ -4859,7 +4859,7 @@ provides-extras = ["proxy", "cli", "extra-proxy", "utils", "caching", "mcp", "sa
 [package.metadata.requires-dev]
 benchmarks = [
    { name = "a2a-sdk", specifier = "==1.1.0" },
-    { name = "mcp", specifier = ">=2.2.0,<3" },
+    { name = "mcp", specifier = ">=2.3.0,<3" },
    { name = "pytest", specifier = "==9.0.3" },
    { name = "pytest-codspeed", specifier = "==4.3.0" },
 ]
@ -4937,7 +4937,7 @@ dev = [
 e2e-dev = [
    { name = "anthropic", specifier = "==0.84.0" },
    { name = "locust", specifier = "==2.45.0" },
-    { name = "mcp", specifier = ">=2.2.0,<3" },
+    { name = "mcp", specifier = ">=2.3.0,<3" },
    { name = "playwright", specifier = "==1.61.0" },
    { name = "psutil", specifier = "==7.2.2" },
    { name = "websockets", specifier = ">=15.0.1,<16.0" },
@ -5415,7 +5415,7 @@ wheels = [

 [[package]]
 name = "mcp"
-version = "2.2.0"
+version = "2.3.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
    { name = "anyio" },
@ -5433,22 +5433,22 @@ dependencies = [
    { name = "typing-inspection" },
    { name = "uvicorn", marker = "sys_platform != 'emscripten'" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/76/31/ac54fb0fdd5b37de704486e288bba4fbbb463f24cfcfedbede407b854513/mcp-2.2.0.tar.gz", hash = "sha256:2dc37ecb1974becdcebdbf7561e7c15a07dbbf20ba21ba16c3593b3038b3afbd", size = 4084129, upload-time = "2026-09-07T16:06:23.439Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/9d/8d/e0d339616f4810e9051d4aba6887afab289ab1f81875fe908b606cdfd0e3/mcp-2.3.0.tar.gz", hash = "sha256:8b147a50441cf059dc88c684e0aeed3687f0aa0f39c6cde7b90330effd2b34d8", size = 4171955, upload-time = "2026-10-02T22:06:56.091Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/1b/ff/8e7eade68b8a28f7da0ed1085544341b51f9c935dbf6b95c76b7edfea6a0/mcp-2.2.0-py3-none-any.whl", hash = "sha256:bde982589473a060ae145e3406e9a5333fe538c97229ba841f5a7f92be004f81", size = 365656, upload-time = "2026-09-07T16:06:19.711Z" },
+    { url = "https://files.pythonhosted.org/packages/20/d7/4a2878f128df729db3ca04e71bfce26405cc2e0c9ebe19c592feeb4c61b9/mcp-2.3.0-py3-none-any.whl", hash = "sha256:dd0c44c089d16453e8ae31a3877a0054d7a2314caaa81f5e0541b9b1734b2377", size = 368664, upload-time = "2026-10-02T22:06:52.119Z" },
 ]

 [[package]]
 name = "mcp-types"
-version = "2.2.0"
+version = "2.3.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
    { name = "pydantic" },
    { name = "typing-extensions" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/ae/91/762d7755d971aff8a28d75f7961656148edf27875c8026e6385aaab08ae7/mcp_types-2.2.0.tar.gz", hash = "sha256:d3ed53703ddd10d9c6399f29d322bb66f3f67ab41348ac8556ba23e07fedefad", size = 65892, upload-time = "2026-09-07T16:06:25.187Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/9e/2d/7c251e34207f6c51000312fc8839111ac45cfe02023f90b44e7f1051dd8e/mcp_types-2.3.0.tar.gz", hash = "sha256:d1e46549edb35ee19a94940fcee6d1addd7e589ab7ea92dda83f5d84781fc362", size = 65898, upload-time = "2026-10-02T22:06:57.863Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/8f/d7/6ffba5d8cd5dd9b8a19478875c50e04945314ba5074e84d749283f27f62d/mcp_types-2.2.0-py3-none-any.whl", hash = "sha256:ea476b73ee86709ab5abc9452385ed36cc05907e582355622e294595c9a04f13", size = 69106, upload-time = "2026-09-07T16:06:21.461Z" },
+    { url = "https://files.pythonhosted.org/packages/ce/48/32583066b155bb5d4507e7887dc88ee23118cc4de8e2a12870c49719e2a8/mcp_types-2.3.0-py3-none-any.whl", hash = "sha256:968efdbdaedfab06adae40d378a34395f1090c5921d4be3c9cde283aaf76d91d", size = 69108, upload-time = "2026-10-02T22:06:54.099Z" },
 ]

 [[package]]