OpenAI Agents SDK - ProvenLog

Prefer zero setup? Use plog run or import provenlog.auto instead. See Auto-Instrumentation.

Setup

from provenlog.integrations.openai_agents import Trail
from agents import add_trace_processor

add_trace_processor(Trail(agent_id="my-openai-agent"))

What gets captured

Event	Action Type	Details
Function calls	`TOOL_CALL` / `TOOL_RESULT`	Function name, arguments, output
LLM generations	`LLM_CALL` / `LLM_RESPONSE`	Model, token counts
Handoffs	`CUSTOM`	Source agent, target agent
Guardrail evaluations	`CUSTOM`	Guardrail name, pass/fail
Agent spans	`CUSTOM`	Agent lifecycle events

How it works

The Trail registers as a trace processor via the OpenAI Agents SDK’s add_trace_processor hook. It receives structured trace events for all agent activity.

Configuration

# Simple — uses default embedded mode
trail = Trail(agent_id="my-openai-agent")

# With explicit client for custom configuration
from provenlog import ProvenLogClient

client = ProvenLogClient("http://localhost:7600", agent_id="my-agent")
trail = Trail(client=client, agent_id="my-openai-agent")