Merge pull request #22188 from BerriAI/litellm_langfuse_key_leakage

fix: langfuse trace leak key on model params

litellm/integrations/langfuse/langfuse.py

@@ -25,6 +25,7 @@ from litellm.litellm_core_utils.core_helpers import (
     reconstruct_model_name,
     filter_exceptions_from_params,
 )
+from litellm.litellm_core_utils.model_param_helper import ModelParamHelper
 from litellm.litellm_core_utils.redact_messages import redact_user_api_key_info
 from litellm.integrations.langfuse.langfuse_mock_client import (
     create_mock_langfuse_client,
@@ -291,8 +292,6 @@ class LangFuseLogger:
 
             functions = optional_params.pop("functions", None)
             tools = optional_params.pop("tools", None)
-            # Remove secret_fields to prevent leaking sensitive data (e.g., authorization headers)
-            optional_params.pop("secret_fields", None)
             if functions is not None:
                 prompt["functions"] = functions
             if tools is not None:
@@ -505,13 +504,18 @@ class LangFuseLogger:
             kwargs.get("model", ""), custom_llm_provider, metadata
         )
 
+        # Use whitelisted model parameters to prevent leaking secrets
+        sanitized_model_params = ModelParamHelper.get_standard_logging_model_parameters(
+            optional_params
+        )
+
         trace.generation(
             CreateGeneration(
                 name=metadata.get("generation_name", "litellm-completion"),
                 startTime=start_time,
                 endTime=end_time,
                 model=model_name,
-                modelParameters=optional_params,
+                modelParameters=sanitized_model_params,
                 prompt=input,
                 completion=output,
                 usage={
@@ -831,13 +835,26 @@ class LangFuseLogger:
                 kwargs.get("model", ""), custom_llm_provider, metadata
             )
 
+            # Use whitelisted model_parameters from StandardLoggingPayload
+            # to prevent leaking secrets (api_key, auth headers, etc.)
+            if standard_logging_object is not None:
+                sanitized_model_params = standard_logging_object.get(
+                    "model_parameters", optional_params
+                )
+            else:
+                sanitized_model_params = (
+                    ModelParamHelper.get_standard_logging_model_parameters(
+                        optional_params
+                    )
+                )
+
             generation_params = {
                 "name": generation_name,
                 "id": clean_metadata.pop("generation_id", generation_id),
                 "start_time": start_time,
                 "end_time": end_time,
                 "model": model_name,
-                "model_parameters": optional_params,
+                "model_parameters": sanitized_model_params,
                 "input": input if not mask_input else "redacted-by-litellm",
                 "output": output if not mask_output else "redacted-by-litellm",
                 "usage": usage,

litellm/litellm_core_utils/litellm_logging.py

@@ -5569,21 +5569,23 @@ def scrub_sensitive_keys_in_metadata(litellm_params: Optional[dict]):
             litellm_params["_langfuse_masking_function"] = masking_fn
         litellm_params["metadata"] = metadata
 
-    ## check user_api_key_metadata for sensitive logging keys
-    cleaned_user_api_key_metadata = {}
-    if "user_api_key_metadata" in metadata and isinstance(
-        metadata["user_api_key_metadata"], dict
-    ):
-        for k, v in metadata["user_api_key_metadata"].items():
-            if k == "logging":  # prevent logging user logging keys
-                cleaned_user_api_key_metadata[k] = (
-                    "scrubbed_by_litellm_for_sensitive_keys"
-                )
-            else:
-                cleaned_user_api_key_metadata[k] = v
+    ## remove sensitive logging/callback keys from metadata dicts
+    ## these contain credentials (langfuse_secret_key, langfuse_public_key, etc.)
+    _sensitive_keys = {"logging", "callback_settings"}
 
-        metadata["user_api_key_metadata"] = cleaned_user_api_key_metadata
-        litellm_params["metadata"] = metadata
+    for metadata_field in (
+        "user_api_key_metadata",
+        "user_api_key_auth_metadata",
+        "user_api_key_team_metadata",
+    ):
+        if metadata_field in metadata and isinstance(metadata[metadata_field], dict):
+            for sensitive_key in _sensitive_keys:
+                metadata[metadata_field].pop(sensitive_key, None)
+
+    ## remove user_api_key_auth entirely - contains full auth object with nested credentials
+    metadata.pop("user_api_key_auth", None)
+
+    litellm_params["metadata"] = metadata
 
     return litellm_params