Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Arize-ai/openinference/llms.txt

Use this file to discover all available pages before exploring further.

OpenInference provides comprehensive Python instrumentation for AI/ML frameworks and LLM providers. All instrumentations are fully OpenTelemetry compatible and can be sent to any OpenTelemetry collector for viewing, such as Arize Phoenix.

​
Available Instrumentations

The following table lists all available Python instrumentation packages:
PackageDescriptionPyPI
openinference-instrumentation-openaiAuto-instrumentation for OpenAI’s Python SDKPyPI
openinference-instrumentation-anthropicAuto-instrumentation for Anthropic SDKPyPI
openinference-instrumentation-langchainAuto-instrumentation for LangChain (1.x and Classic)PyPI
openinference-instrumentation-llama-indexAuto-instrumentation for LlamaIndexPyPI
openinference-instrumentation-bedrockAuto-instrumentation for AWS Bedrock (boto3 and aioboto3)PyPI
openinference-instrumentation-vertexaiAuto-instrumentation for Google Vertex AIPyPI
openinference-instrumentation-mistralaiAuto-instrumentation for Mistral AI SDKPyPI
openinference-instrumentation-dspyAuto-instrumentation for DSPyPyPI
openinference-instrumentation-haystackAuto-instrumentation for Haystack pipelinesPyPI
openinference-instrumentation-groqAuto-instrumentation for Groq SDKPyPI
openinference-instrumentation-litellmAuto-instrumentation for LiteLLMPyPI
openinference-instrumentation-instructorAuto-instrumentation for Instructor libraryPyPI
openinference-instrumentation-crewaiAuto-instrumentation for CrewAI agentsPyPI
openinference-instrumentation-google-genaiAuto-instrumentation for Google GenAIPyPI
openinference-instrumentation-google-adkAuto-instrumentation for Google ADKPyPI
openinference-instrumentation-pydantic-aiAuto-instrumentation for PydanticAIPyPI
openinference-instrumentation-openai-agentsAuto-instrumentation for OpenAI Agents SDKPyPI
openinference-instrumentation-autogenAuto-instrumentation for Autogen (ag2)PyPI
openinference-instrumentation-autogen-agentchatAuto-instrumentation for Autogen AgentChatPyPI
openinference-instrumentation-agent-frameworkAuto-instrumentation for Microsoft Agent FrameworkPyPI
openinference-instrumentation-agentspecAuto-instrumentation for Agent SpecPyPI
openinference-instrumentation-agnoAuto-instrumentation for Agno agentsPyPI
openinference-instrumentation-smolagentsAuto-instrumentation for smolagentsPyPI
openinference-instrumentation-strands-agentsAuto-instrumentation for Strands AgentsPyPI
openinference-instrumentation-beeaiAuto-instrumentation for BeeAIPyPI
openinference-instrumentation-guardrailsAuto-instrumentation for GuardrailsPyPI
openinference-instrumentation-openllmetryConverts OpenLLMetry traces to OpenInferencePyPI
openinference-instrumentation-openlitAuto-instrumentation for OpenLitPyPI
openinference-instrumentation-pipecatAuto-instrumentation for PipecatPyPI
openinference-instrumentation-portkeyAuto-instrumentation for PortkeyPyPI
openinference-instrumentation-mcpAuto-instrumentation for MCP SDK (context propagation)PyPI
openinference-instrumentation-promptflowAuto-instrumentation for PromptFlowPyPI

​
Getting Started

All instrumentations follow a similar pattern:
  1. Install the instrumentation package
  2. Configure the OpenTelemetry tracer provider
  3. Instrument your application
  4. Run your code and view traces

​
Basic Example

from openinference.instrumentation.<package> import <Package>Instrumentor
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk import trace as trace_sdk
from opentelemetry.sdk.trace.export import SimpleSpanProcessor

# Configure tracer
endpoint = "http://127.0.0.1:6006/v1/traces"
tracer_provider = trace_sdk.TracerProvider()
tracer_provider.add_span_processor(SimpleSpanProcessor(OTLPSpanExporter(endpoint)))

# Instrument
<Package>Instrumentor().instrument(tracer_provider=tracer_provider)

# Your application code here

​
Common Features

All OpenInference instrumentations support:
  • Session tracking - Track user sessions across multiple requests
  • User identification - Associate traces with specific users
  • Metadata and tags - Add custom metadata and tags to traces
  • Prompt template tracking - Track prompt templates and variables
  • PII masking - Configure trace masking for sensitive data
  • Context propagation - Automatic propagation of tracing context

​
Using Context Attributes

from openinference.instrumentation import using_attributes

with using_attributes(
    session_id="my-session-id",
    user_id="user-123",
    metadata={"key": "value"},
    tags=["production", "feature-x"],
):
    # Your instrumented code here
    pass

​
Additional Resources

​
Contributing

To contribute a new instrumentation or report issues, visit the OpenInference GitHub repository.