Add Langfuse observability to Unified API

backend/app/core/langfuse/langfuse.py

@@ -1,10 +1,12 @@
 import uuid
 import logging
-from typing import Any, Dict, Optional
+from typing import Any, Callable, Dict, Optional
+from functools import wraps
 
 from asgi_correlation_id import correlation_id
 from langfuse import Langfuse
 from langfuse.client import StatefulGenerationClient, StatefulTraceClient
+from app.models.llm import CompletionConfig, QueryParams, LLMCallResponse
 
 logger = logging.getLogger(__name__)
 
@@ -107,3 +109,112 @@ def log_error(self, error_message: str, response_id: Optional[str] = None):
 
     def flush(self):
         self.langfuse.flush()
+
+
+def observe_llm_execution(
+    session_id: str | None = None,
+    credentials: dict | None = None,
+):
+    """Decorator to add Langfuse observability to LLM provider execute methods.
+
+    Args:
+        credentials: Langfuse credentials with public_key, secret_key, and host
+        session_id: Session ID for grouping traces (conversation_id)
+
+    Usage:
+        decorated_execute = observe_llm_execution(
+            credentials=langfuse_creds,
+            session_id=conversation_id
+        )(provider_instance.execute)
+    """
+
+    def decorator(func: Callable) -> Callable:
+        @wraps(func)
+        def wrapper(completion_config: CompletionConfig, query: QueryParams, **kwargs):
+            # Skip observability if no credentials provided
+            if not credentials or not all(
+                key in credentials for key in ["public_key", "secret_key", "host"]
+            ):
+                logger.info("[Langfuse] No credentials - skipping observability")
+                return func(completion_config, query, **kwargs)
+
+            try:
+                langfuse = Langfuse(
+                    public_key=credentials.get("public_key"),
+                    secret_key=credentials.get("secret_key"),
+                    host=credentials.get("host"),
+                )
+            except Exception as e:
+                logger.warning(f"[Langfuse] Failed to initialize client: {e}")
+                return func(completion_config, query, **kwargs)
+
+            trace_metadata = {
+                "provider": completion_config.provider,
+            }
+
+            if query.conversation and query.conversation.id:
+                trace_metadata["conversation_id"] = query.conversation.id
+
+            trace = langfuse.trace(
+                name="unified-llm-call",
+                input=query.input,
+                metadata=trace_metadata,
+                tags=[completion_config.provider],
+            )
+
+            generation = trace.generation(
+                name=f"{completion_config.provider}-completion",
+                input=query.input,
+                model=completion_config.params.get("model"),
+            )
+
+            try:
+                # Execute the actual LLM call
+                response: LLMCallResponse | None
+                error: str | None
+                response, error = func(completion_config, query, **kwargs)
+
+                if response:
+                    generation.end(
+                        output={
+                            "status": "success",
+                            "output": response.response.output.text,
+                        },
+                        usage_details={
+                            "input": response.usage.input_tokens,
+                            "output": response.usage.output_tokens,
+                        },
+                        model=response.response.model,
+                    )
+
+                    trace.update(
+                        output={
+                            "status": "success",
+                            "output": response.response.output.text,
+                        },
+                        session_id=session_id or response.response.conversation_id,
+                    )
+                else:
+                    error_msg = error or "Unknown error"
+                    generation.end(output={"error": error_msg})
+                    trace.update(
+                        output={"status": "failure", "error": error_msg},
+                        session_id=session_id,
+                    )
+
+                langfuse.flush()
+                return response, error
+
+            except Exception as e:
+                error_msg = str(e)
+                generation.end(output={"error": error_msg})
+                trace.update(
+                    output={"status": "failure", "error": error_msg},
+                    session_id=session_id,
+                )
+                langfuse.flush()
+                raise
+
+        return wrapper
+
+    return decorator

backend/app/services/llm/jobs.py

@@ -7,11 +7,13 @@
 
 from app.core.db import engine
 from app.crud.config import ConfigVersionCrud
+from app.crud.credentials import get_provider_credential
 from app.crud.jobs import JobCrud
 from app.models import JobStatus, JobType, JobUpdate, LLMCallRequest
 from app.models.llm.request import ConfigBlob, LLMCallConfig
 from app.utils import APIResponse, send_callback
 from app.celery.utils import start_high_priority_job
+from app.core.langfuse.langfuse import observe_llm_execution
 from app.services.llm.providers.registry import get_llm_provider
 
 
@@ -182,7 +184,25 @@ def execute_job(
                 )
                 return handle_job_error(job_id, request.callback_url, callback_response)
 
-        response, error = provider_instance.execute(
+            langfuse_credentials = get_provider_credential(
+                session=session,
+                org_id=organization_id,
+                project_id=project_id,
+                provider="langfuse",
+            )
+
+        # Extract conversation_id for langfuse session grouping
+        conversation_id = None
+        if request.query.conversation and request.query.conversation.id:
+            conversation_id = request.query.conversation.id
+
+        # Apply Langfuse observability decorator to provider execute method
+        decorated_execute = observe_llm_execution(
+            credentials=langfuse_credentials,
+            session_id=conversation_id,
+        )(provider_instance.execute)
+
+        response, error = decorated_execute(
             completion_config=config_blob.completion,
             query=request.query,
             include_provider_raw_response=request.include_provider_raw_response,