diff --git a/litellm/proxy/guardrails/guardrail_hooks/custom_code/custom_code_guardrail.py b/litellm/proxy/guardrails/guardrail_hooks/custom_code/custom_code_guardrail.py index d5ef1e949b8..e0291975699 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/custom_code/custom_code_guardrail.py +++ b/litellm/proxy/guardrails/guardrail_hooks/custom_code/custom_code_guardrail.py @@ -58,6 +58,11 @@ if TYPE_CHECKING: from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +def _metadata_bucket(request_data: Mapping[str, object], key: str) -> Mapping[str, object]: + bucket: Final = request_data.get(key) + return bucket if isinstance(bucket, Mapping) else {} + + class CustomCodeGuardrailError(Exception): """Raised when custom code guardrail execution fails.""" @@ -280,12 +285,16 @@ class CustomCodeGuardrail(CustomGuardrail): Returns: Safe subset of request data """ + metadata: Final = { + **_metadata_bucket(request_data, "metadata"), + **_metadata_bucket(request_data, "litellm_metadata"), + } return { "model": request_data.get("model"), - "user_id": request_data.get("user_api_key_user_id"), - "team_id": request_data.get("user_api_key_team_id"), - "end_user_id": request_data.get("user_api_key_end_user_id"), - "metadata": request_data.get("metadata", {}), + "user_id": metadata.get("user_api_key_user_id"), + "team_id": metadata.get("user_api_key_team_id"), + "end_user_id": metadata.get("user_api_key_end_user_id"), + "metadata": metadata, } def _process_result( diff --git a/litellm/proxy/utils.py b/litellm/proxy/utils.py index 479bd0a55af..b6c487c0edf 100644 --- a/litellm/proxy/utils.py +++ b/litellm/proxy/utils.py @@ -1269,7 +1269,12 @@ class ProxyLogging: # (e.g. MCPJWTSigner) to independently verify the caller's identity # before re-signing an outbound token (FR-5 verify+re-sign). "incoming_bearer_token": kwargs.get("incoming_bearer_token"), - "metadata": {"headers": kwargs.get("headers") or {}}, + "metadata": { + "headers": kwargs.get("headers") or {}, + "user_api_key_user_id": kwargs.get("user_api_key_user_id"), + "user_api_key_team_id": kwargs.get("user_api_key_team_id"), + "user_api_key_end_user_id": kwargs.get("user_api_key_end_user_id"), + }, } user_api_key_auth: Final = kwargs.get("user_api_key_auth") if isinstance(user_api_key_auth, UserAPIKeyAuth): diff --git a/tests/test_litellm/proxy/guardrails/test_custom_code_security.py b/tests/test_litellm/proxy/guardrails/test_custom_code_security.py index 7971cf62c9a..068cd0d8ed7 100644 --- a/tests/test_litellm/proxy/guardrails/test_custom_code_security.py +++ b/tests/test_litellm/proxy/guardrails/test_custom_code_security.py @@ -250,6 +250,76 @@ async def test_custom_code_flag_default_reason_and_empty_metadata(): } +IDENTITY_ECHO_CODE = ( + "def apply_guardrail(inputs, request_data, input_type):\n" + " return flag('identity', metadata={\n" + " 'ids': [request_data['user_id'], request_data['team_id'], request_data['end_user_id']],\n" + " 'metadata_keys': sorted(request_data['metadata'].keys()),\n" + " })\n" +) +CALLER_IDENTITY = { + "user_api_key_user_id": "someone@example.com", + "user_api_key_team_id": "team-1", + "user_api_key_end_user_id": "end-user-1", + "user_api_key_alias": "guardrail-repro-key", +} + + +@pytest.mark.asyncio +@pytest.mark.parametrize("metadata_key", ["metadata", "litellm_metadata"]) +async def test_custom_code_sandbox_sees_caller_identity_from_proxy_metadata_bucket(metadata_key): + """LIT-6609: the proxy writes user_api_key_* into `metadata` (chat) or `litellm_metadata` + (/v1/messages, responses, batches, files); the sandbox must resolve ids from either.""" + guardrail = _compile(IDENTITY_ECHO_CODE) + request_data = {"model": "m", metadata_key: dict(CALLER_IDENTITY)} + + await guardrail.apply_guardrail(inputs={"texts": ["x"]}, request_data=request_data, input_type="request") + + entry = request_data[metadata_key]["standard_logging_guardrail_information"][0] + assert entry["guardrail_response"]["metadata"] == { + "ids": ["someone@example.com", "team-1", "end-user-1"], + "metadata_keys": sorted(CALLER_IDENTITY), + } + + +@pytest.mark.asyncio +async def test_custom_code_sandbox_merges_caller_metadata_with_litellm_metadata(): + """On litellm_metadata routes the caller's own `metadata` field must stay visible next to + the proxy identity block, and the proxy block wins on key collisions.""" + guardrail = _compile(IDENTITY_ECHO_CODE) + request_data = { + "model": "m", + "metadata": {"trace_id": "abc", "user_api_key_user_id": "forged"}, + "litellm_metadata": dict(CALLER_IDENTITY), + } + + await guardrail.apply_guardrail(inputs={"texts": ["x"]}, request_data=request_data, input_type="request") + + entry = request_data["litellm_metadata"]["standard_logging_guardrail_information"][0] + assert entry["guardrail_response"]["metadata"] == { + "ids": ["someone@example.com", "team-1", "end-user-1"], + "metadata_keys": sorted([*CALLER_IDENTITY, "trace_id"]), + } + + +@pytest.mark.asyncio +async def test_custom_code_sandbox_ignores_top_level_identity_fields(): + """Only the proxy-owned metadata buckets carry identity; user_api_key_* keys at the top level + of the request body are caller-controlled on ordinary routes and must never become ids.""" + code = ( + "def apply_guardrail(inputs, request_data, input_type):\n" + " ids = [request_data['user_id'], request_data['team_id'], request_data['end_user_id']]\n" + " return flag('identity', metadata={'ids': str(ids)})\n" + ) + guardrail = _compile(code) + request_data = {"model": "m", **CALLER_IDENTITY, "metadata": {"headers": {}}} + + await guardrail.apply_guardrail(inputs={"texts": ["x"]}, request_data=request_data, input_type="request") + + entry = request_data["metadata"]["standard_logging_guardrail_information"][0] + assert entry["guardrail_response"]["metadata"]["ids"] == "[None, None, None]" + + @pytest.mark.asyncio async def test_custom_code_allow_still_records_success_not_flagged(): code = "def apply_guardrail(inputs, request_data, input_type):\n return allow()\n" diff --git a/tests/test_litellm/proxy/utils/proxy_logging/test_mcp_bridging.py b/tests/test_litellm/proxy/utils/proxy_logging/test_mcp_bridging.py index 56057dce7e0..438b2351034 100644 --- a/tests/test_litellm/proxy/utils/proxy_logging/test_mcp_bridging.py +++ b/tests/test_litellm/proxy/utils/proxy_logging/test_mcp_bridging.py @@ -87,6 +87,27 @@ def test_convert_mcp_to_llm_format_exposes_headers_on_metadata(proxy_logging, ma assert out["metadata"]["headers"] == {"x-nuid": "nuid-1"} +def test_convert_mcp_to_llm_format_exposes_caller_identity_on_metadata(proxy_logging, make_mcp_request_obj): + """Custom code guardrails resolve user_id/team_id/end_user_id from the proxy-owned metadata + bucket on every route, so the MCP bridge has to write the authenticated ids there too.""" + req = make_mcp_request_obj() + out = proxy_logging._convert_mcp_to_llm_format( + request_obj=req, + kwargs={ + "user_api_key_user_id": "u-1", + "user_api_key_team_id": "t-1", + "user_api_key_end_user_id": "eu-1", + "headers": {"x-nuid": "nuid-1"}, + }, + ) + assert out["metadata"] == { + "headers": {"x-nuid": "nuid-1"}, + "user_api_key_user_id": "u-1", + "user_api_key_team_id": "t-1", + "user_api_key_end_user_id": "eu-1", + } + + def test_convert_mcp_to_llm_format_defaults_headers_to_empty(proxy_logging, make_mcp_request_obj): req = make_mcp_request_obj() out = proxy_logging._convert_mcp_to_llm_format(request_obj=req, kwargs={})