Implementation:Truera Trulens TruChain Wrapper

Knowledge Sources	TruLens TruLens LangChain Guide
Domains	Observability, LLM_Evaluation
Last Updated	2026-02-14 08:00 GMT

Overview

Concrete tool for wrapping and instrumenting LangChain Runnable applications provided by the trulens-apps-langchain package.

Description

The TruChain class is a recorder for LangChain applications. It wraps a LangChain Runnable (chain, agent, or pipeline) and automatically instruments its components for tracing and evaluation. TruChain discovers LangChain components through the Runnable's internal structure and applies OTEL instrumentation to capture spans for retrievers, LLMs, and output parsers.

TruChain inherits from the base App class and adds LangChain-specific component discovery and instrumentation logic.

Usage

Import and use TruChain when you have a LangChain application that you want to evaluate. Wrap the chain after defining feedback functions. Use the context manager pattern to record traces.

Code Reference

Source Location

Repository: trulens
File: src/apps/langchain/trulens/apps/langchain/tru_chain.py
Lines: L584-885

Signature

class TruChain(core_app.App):
    app: Runnable
    """The langchain app to be instrumented."""

    def __init__(
        self,
        app: Runnable,
        main_method: Optional[Callable] = None,
        **kwargs: Dict[str, Any],
    ):
        """
        Args:
            app: A LangChain Runnable (chain, agent, pipeline).
            main_method: Optional callable for the main entry point.
            **kwargs: Additional arguments including:
                app_name (str): Application name for identification.
                app_version (str): Version tag.
                feedbacks (List[Feedback]): Feedback functions for evaluation.
                connector (DBConnector): Optional database connector.
        """

Import

from trulens.apps.langchain import TruChain

I/O Contract

Inputs

Name	Type	Required	Description
app	Runnable	Yes	LangChain Runnable (chain, agent, pipeline)
main_method	Callable	No	Optional entry point method
app_name	str	No	Application name (via kwargs)
app_version	str	No	Version tag (via kwargs)
feedbacks	List[Feedback]	No	Feedback functions for evaluation (via kwargs)

Outputs

Name	Type	Description
return	TruChain	Instrumented wrapper supporting context manager recording

Usage Examples

Basic LangChain RAG Wrapping

from trulens.apps.langchain import TruChain
from trulens.core.session import TruSession

session = TruSession()

# Assuming rag_chain is a LangChain Runnable
tru_recorder = TruChain(
    rag_chain,
    app_name="RAG_App",
    app_version="v1",
    feedbacks=[f_answer_relevance, f_context_relevance, f_groundedness],
)

# Record a trace
with tru_recorder as recording:
    result = rag_chain.invoke("What is LangChain?")

# Access recording
record = recording.get()
print(record.main_output)

Related Pages

Implements Principle

Principle:Truera_Trulens_LangChain_App_Wrapping

Requires Environment

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment