@respan/instrumentation-llama-index

Respan instrumentation plugin for LlamaIndex TypeScript.

The instrumentor attaches to LlamaIndex's callback manager and emits Respan OTEL spans for LLM calls, query engines, retrievers, synthesizers, agents, and tool executions.

Install

npm install @respan/respan @respan/instrumentation-llama-index llamaindex @llamaindex/openai

Usage

import { Respan } from "@respan/respan";
import { LlamaIndexInstrumentor } from "@respan/instrumentation-llama-index";
import { OpenAI } from "@llamaindex/openai";
import * as LlamaIndex from "llamaindex";

const workflowName = "llama_index_ts_basic_llm";

const respan = new Respan({
  appName: workflowName,
  traceContent: true,
  instrumentations: [
    new LlamaIndexInstrumentor({
      workflowName,
      llamaIndexModule: LlamaIndex,
    }),
  ],
});
await respan.initialize();

try {
  await respan.propagateAttributes(
    {
      trace_group_identifier: workflowName,
      metadata: { workflow_name: workflowName },
    },
    () =>
      respan.withWorkflow({ name: workflowName }, async () => {
        const llm = new OpenAI({ model: "gpt-4.1-nano" });
        const response = await llm.complete({
          prompt: "Say hello in three languages.",
        });
        console.log(response.text);
      }),
  );
} finally {
  await respan.shutdown();
}

Use workflowName when you want auto-emitted LlamaIndex spans to carry the same workflow label that appears in Respan traces. Use the same value as the trace_group_identifier when examples or application requests should be found by workflow name.