Supported frameworks and libraries

OpenLLMetry automatically instruments the following provider APIs and libraries:

OpenAI
Azure OpenAI
Anthropic
Cohere
Bedrock (AWS)
Vertex AI (GCP)

Supported vector databases

The following

Chroma
Pinecone
Qdrant

Supported frameworks

The following LLM frameworks are supported:

LangChain
LlamaIndex

Install and configure the Node.js instrumentation

Perform the following tasks to configure Node.js instrumentation:

Add traceloop SDK as a dependency:

npm install --save @traceloop/node-server-sdk

Initialize the SDK:

traceloop.initialize(  
    appName: "<YOUR_SERVICE_NAME>",
  );

Set environment variables

Set the following environment variables:

Environment variable	Example Values	Description	Optional?
`TRACELOOP_BASE_URL`	`http://<YOUR_OBSERVE_AGENT_HOSTNAME>:4318`	OTLP endpoint	No
`TRACELOOP_TRACE_CONTENT`	`true` / `false`	Enables or disables extraction of inputs and outputs from LLM calls	Yes
`OTEL_RESOURCE_ATTRIBUTES`	`deployment.environment=dev,service.namespace=inference`	A list of key=value resource attributes you wish to add to your spans	Yes

Add attributes at runtime

To annotate your span data with custom attributes like customer_id, user_id, and so on, we recommend using OpenLLMetry functions. Depending on your use-case, you may choose one of the following approaches:

For class methods

/* This example wraps a workflow span as denoted by the "workflow" decorator. To wrap spans of other types you may use the task/agent/tool decorators as desired  
  */
  class MyClass {
    @traceloop.workflow({ associationProperties: { userId: "user123" })
    myMethod() {
      // Your code here
    }
  }

Within attributes per workflow / task / agent / tool call spans

// Wraps workflows  
  // Replace with withTask / withAgent / withTool as needed
  traceloop.withWorkflow(
    \{
      name: "workflow_name",
      associationProperties: { userId: "user12345", chatId: "chat12345" },
    },
    () => {
      // Your code here
      // (function can be made async if needed)
    }
  );

Directly wrapping with global attributes

traceloop.withAssociationProperties(  
    \{
      userId: "user12345",
      chatId: "chat12345",
    },
    () => {
      // Your code here
      // (can be async or sync)
    }
  );