introspection-sdk 0.4.0

Python SDK for Introspection — continuously improve your AI systems with production feedback and frontier practices

Build frontier AI systems that self-improve.

Introspection continuously improves your AI systems with production feedback and frontier practices. This is the Python SDK.

Install

uv add introspection-sdk
# or
pip install introspection-sdk

Optional Extras

pip install 'introspection-sdk[openai-agents]'  # OpenAI Agents SDK
pip install 'introspection-sdk[langfuse]'        # Langfuse
pip install 'introspection-sdk[braintrust]'      # Braintrust
pip install 'introspection-sdk[arize]'           # Arize Phoenix + OpenInference

Environment Variables

export INTROSPECTION_TOKEN="intro_xxx"
export INTROSPECTION_BASE_URL="https://otel.introspection.dev"  # optional

Quickstart

OpenTelemetry Span Processor

from introspection_sdk import IntrospectionSpanProcessor
import logfire

logfire.configure(
    additional_span_processors=[IntrospectionSpanProcessor()],
)

logfire.instrument_openai()

OpenAI Agents SDK

from agents import Agent, Runner, set_trace_processors, tool
from introspection_sdk import IntrospectionTracingProcessor

set_trace_processors([IntrospectionTracingProcessor()])

@tool
def get_weather(city: str) -> str:
    """Get the current weather for a city."""
    return f"Sunny, 72F in {city}"

agent = Agent(name="Weather Bot", tools=[get_weather])
result = await Runner.run(agent, "What's the weather in Tokyo?")

Reasoning Model Support

Some models produce reasoning items that the OpenAI Conversations API rejects. IntrospectionConversationsSession strips those items transparently:

from introspection_sdk import IntrospectionConversationsSession

session = IntrospectionConversationsSession(conversation_id="conv_123")
result = await Runner.run(agent, "Hello!", session=session)

Claude Agent SDK

from introspection_sdk import ClaudeTracingProcessor

processor = ClaudeTracingProcessor()
processor.configure()

# All ClaudeSDKClient instances are now automatically traced

LangChain / LangGraph

from introspection_sdk import IntrospectionCallbackHandler

handler = IntrospectionCallbackHandler(service_name="my-app")
response = model.invoke("Hello!", config={"callbacks": [handler]})

Anthropic SDK

from introspection_sdk.anthropic import AnthropicInstrumentor

instrumentor = AnthropicInstrumentor()
instrumentor.instrument(tracer_provider=provider)

# All client.messages.create calls are traced, including thinking blocks

OpenInference (Arize, Langfuse, Braintrust)

from opentelemetry.sdk.trace import TracerProvider
from openinference.instrumentation.openai import OpenAIInstrumentor
from introspection_sdk import IntrospectionSpanProcessor

provider = TracerProvider()
provider.add_span_processor(IntrospectionSpanProcessor())
OpenAIInstrumentor().instrument(tracer_provider=provider)

See examples/ for complete integration patterns including dual-export with Arize, Langfuse, Braintrust, and LangSmith.

Client API

from introspection_sdk import IntrospectionClient

client = IntrospectionClient()

with client.set_user_id("user_123"):
    with client.set_conversation("conv_456", previous_response_id="msg_123"):
        client.feedback("thumbs_up", comments="Great response!")

client.shutdown()

Methods

Method	Description
feedback(type, **kwargs)	Track feedback on AI responses
identify(user_id, traits=)	Associate a user with traits (context manager)
track(event, properties=)	Track any user action
flush(timeout_ms=30000)	Flush pending events
shutdown()	Shutdown and flush

Context Managers

Method	Description
set_user_id(id)	Set user context
set_conversation(id?, response_id?)	Set conversation context
set_agent(name, id?)	Set agent context
set_anonymous_id(id)	Set anonymous ID
set_baggage(**values)	Set arbitrary baggage values

Documentation

Full documentation is available at docs.introspection.dev.