Implementation:Confident ai Deepeval Update Current Trace

Overview

Update Current Trace is the implementation function that enriches the currently active trace with metadata for filtering, grouping, and analysis in the Confident AI dashboard. It allows developers to inject contextual information -- such as trace names, tags, user identifiers, and custom metadata -- into traces from within any observed function.

API Documentation

Function: update_current_trace

Source: deepeval/tracing/context.py:L62-118

Import:

from deepeval.tracing import update_current_trace

Signature:

update_current_trace(
    name=None,
    tags=None,
    metadata=None,
    thread_id=None,
    user_id=None,
    input=None,
    output=None,
    test_case=None,
    confident_api_key=None,
    test_case_id=None,
)

Parameters

Parameter	Type	Description
name	Optional[str]	A human-readable name for the trace (e.g., "customer-support-agent").
tags	Optional[List[str]]	A list of string tags for categorization and filtering (e.g., ["production", "v2"]).
metadata	Optional[Dict]	Arbitrary key-value pairs for custom contextual data (e.g., {"model_version": "gpt-4o"}).
thread_id	Optional[str]	Identifier for the conversation thread, enabling multi-turn conversation tracking.
user_id	Optional[str]	Identifier for the user who triggered the trace.
input	Optional	Override the automatically captured trace input.
output	Optional	Override the automatically captured trace output.
test_case	Optional	Associate a test case with the trace.
confident_api_key	Optional[str]	Override the default Confident AI API key for this trace.
test_case_id	Optional[str]	Associate a specific test case ID with the trace.

Input / Output

• Inputs: Metadata values (name, tags, metadata dict, thread_id, user_id, and optional overrides).
• Outputs: The active trace is enriched with the provided metadata, which appears in the Confident AI dashboard for filtering and analysis.

Usage Example

from deepeval.tracing import observe, update_current_trace

@observe(type="agent")
def my_agent(query: str) -> str:
    update_current_trace(
        name="customer-support-agent",
        tags=["production", "v2"],
        metadata={"model_version": "gpt-4o"},
        thread_id="conv-123",
        user_id="user-456",
    )
    return process(query)

Relationships

Principle:Confident_ai_Deepeval_Trace_Metadata_Enrichment

Metadata

DeepEval Tracing Observability LLM_Evaluation 2026-02-14 09:00 GMT