OpenInference Agent Spec Instrumentation

Python auto-instrumentation library for Agent Spec.

Open Agent Spec (Agent Spec) is a portable, platform-agnostic configuration language that allows Agents and Agentic Systems to be described with high fidelity. It defines the conceptual building blocks—called components—that make up agents in typical agent-based systems. This includes the properties that configure each component and the semantics that govern their behavior.

Agent Spec Tracing is an extension of Agent Spec that standardizes how agent and flow executions emit traces. It enables:

• Runtime adapters to emit consistent traces across different frameworks.
• Consumers (observability backends, UIs, developer tooling) to ingest one standardized format regardless of the producer.

You can find more information about Agent Spec and Agent Spec Tracing at:

• Agent Spec: https://oracle.github.io/agent-spec/development/agentspec/index.html
• Tracing: https://oracle.github.io/agent-spec/development/agentspec/tracing.html

The traces emitted by this instrumentation are fully OpenTelemetry compatible and can be sent to an OpenTelemetry collector for viewing, such as Arize Phoenix or Arize AX.

Installation

pip install openinference-instrumentation-agentspec

Quickstart

In this example we will instrument a small program that uses Agent Spec Tracing and observe the traces via arize-phoenix.

Install packages.

pip install openinference-instrumentation-agentspec pyagentspec[langgraph] arize-phoenix opentelemetry-sdk opentelemetry-exporter-otlp

Start the phoenix server so that it is ready to collect traces. The Phoenix server runs entirely on your machine and does not send data over the internet.

phoenix serve

In a python file (e.g., agentspec_agent.py) , set up the AgentSpecInstrumentor and configure the tracer to send traces to Phoenix.

from pyagentspec.adapters.langgraph import AgentSpecLoader
from pyagentspec.agent import Agent
from pyagentspec.llms import OpenAiConfig

agent = Agent(
    name="assistant",
    description="An general purpose agent without tools",
    llm_config=OpenAiConfig(name="openai-gpt-5-mini", model_id="gpt-5-mini"),
    system_prompt="You are a helpful assistant. Help the user answering politely.",
)

langgraph_agent = AgentSpecLoader().load_component(agent)

from phoenix.otel import register

from openinference.instrumentation.agentspec import AgentSpecInstrumentor

tracer_provider = register(batch=True, project_name="hello-world-app")
AgentSpecInstrumentor().instrument(tracer_provider=tracer_provider)

while True:
    user_input = input("USER  >>> ")
    if user_input.lower() in ["exit", "quit"]:
        break
    response = langgraph_agent.invoke(
        input={"messages": [{"role": "user", "content": user_input}]},
        config={"configurable": {"thread_id": "1"}},
    )
    print("AGENT >>>", response['messages'][-1].content.strip())

Now simply run the python file and observe the traces in Phoenix.

python agentspec_agent.py

More Info

• More info on OpenInference and Phoenix
• More info on OpenInference and Arize AX
• How to customize spans to track sessions, metadata, etc.
• How to account for private information and span payload customization