OpenInference Instrumentor

OpenInferenceInstrumentor is a generic wrapper that adapts any OpenInference instrumentor to Respan’s plugin protocol. This lets you use the full OpenInference ecosystem (30+ instrumentors) with Respan.

Install

$
pip install respan-ai respan-instrumentation-openinference

Then install the specific OpenInference instrumentor you need:

$
pip install openinference-instrumentation-crewai      # CrewAI
$
pip install openinference-instrumentation-langchain    # LangChain
$
pip install openinference-instrumentation-llama-index  # LlamaIndex
$
pip install openinference-instrumentation-openai       # OpenAI
$
pip install openinference-instrumentation-anthropic    # Anthropic
$
pip install openinference-instrumentation-bedrock      # AWS Bedrock
$
pip install openinference-instrumentation-google-adk   # Google ADK
$
pip install openinference-instrumentation-pydantic-ai  # Pydantic AI
$
# ... and 20+ more

Usage

1
from respan import Respan
2
from respan_instrumentation_openinference import OpenInferenceInstrumentor
3
4
# Pass any OpenInference instrumentor class
5
from openinference.instrumentation.crewai import CrewAIInstrumentor
6
7
respan = Respan(
8
    instrumentations=[
9
        OpenInferenceInstrumentor(instrumentor=CrewAIInstrumentor())
10
    ]
11
)
12
13
# All CrewAI calls are now auto-traced and sent to Respan

How it works

1. The wrapper calls instrumentor.instrument() which patches the target library
2. OpenInference spans are emitted via OpenTelemetry
3. Respan’s composite processor auto-enriches OpenInference attributes into Traceloop/GenAI equivalents
4. Spans appear in Respan with full model, tokens, input/output data

Supported instrumentors

See the full OpenInference list — any instrumentor that follows the OpenInference protocol works with this wrapper.