openinference

OpenInference Instrumentation for TypeSafe AI SDK

This module provides automatic instrumentation for the TypeSafe AI Node.js SDK (@typesafe-ai/sdk), which may be used in conjunction with @opentelemetry/sdk-trace-node.

Each TypeSafeClient.systemOne call produces one OpenInference LLM span with JSON input.value / output.value, model, and token usage. client.models.list() is not instrumented. Traces follow the OpenInference semantic conventions and can be sent to any OpenTelemetry-compatible backend such as Arize Phoenix or Arize AX.

Requires Node.js 20+.

Installation

npm install --save @arizeai/openinference-instrumentation-typesafe @typesafe-ai/sdk

Usage

import { NodeTracerProvider } from "@opentelemetry/sdk-trace-node";
import { registerInstrumentations } from "@opentelemetry/instrumentation";
import { TypeSafeInstrumentation } from "@arizeai/openinference-instrumentation-typesafe";

const provider = new NodeTracerProvider();
provider.register();

registerInstrumentations({
  instrumentations: [new TypeSafeInstrumentation()],
});

For ESM, bundlers, or when the SDK is imported first, call manuallyInstrument:

import * as TypeSafe from "@typesafe-ai/sdk";
import { TypeSafeInstrumentation } from "@arizeai/openinference-instrumentation-typesafe";

const instrumentation = new TypeSafeInstrumentation();
instrumentation.manuallyInstrument(TypeSafe);

const client = new TypeSafe.TypeSafeClient();
const { data, requestId } = await client
  .systemOne({
    state: { document: "I was charged twice. Please fix this ASAP." },
    questions: {
      category: TypeSafe.choice("What is this ticket about?", {
        billing: null,
        technical: null,
        other: null,
      }),
    },
  })
  .withResponse();

console.log(data.answers.category, requestId);

The wrapped return value remains an SDK APIPromise (await, withResponse(), map(), etc.).

Configuration

The TypeSafeInstrumentation constructor accepts:

instrumentationConfig — standard OpenTelemetry instrumentation configuration
traceConfig — OpenInference masking/redaction options (hideInputs, hideOutputs, …)
tracerProvider — optional custom tracer provider (defaults to the global provider)

const instrumentation = new TypeSafeInstrumentation({
  tracerProvider,
  instrumentationConfig: { enabled: true },
  traceConfig: {
    hideInputs: true,
    hideOutputs: true,
  },
});

Examples

Runnable examples live in the repository: openinference-instrumentation-typesafe/examples.

For more information on OpenTelemetry Node.js SDK, see the OpenTelemetry Node.js SDK documentation.

Using a Custom Tracer Provider

You can specify a custom tracer provider when creating the instrumentation. This is useful when you want to use a non-global tracer provider or have more control over the tracing configuration.

import { NodeTracerProvider } from "@opentelemetry/sdk-trace-node";
import { resourceFromAttributes } from "@opentelemetry/resources";
import { registerInstrumentations } from "@opentelemetry/instrumentation";
import { SEMRESATTRS_PROJECT_NAME } from "@arizeai/openinference-semantic-conventions";
import { TypeSafeInstrumentation } from "@arizeai/openinference-instrumentation-typesafe";

const customTracerProvider = new NodeTracerProvider({
  resource: resourceFromAttributes({
    [SEMRESATTRS_PROJECT_NAME]: "my-typesafe-project",
  }),
});

const instrumentation = new TypeSafeInstrumentation({
  tracerProvider: customTracerProvider,
});

registerInstrumentations({
  instrumentations: [instrumentation],
  tracerProvider: customTracerProvider,
});

Alternatively, set the tracer provider after creating the instrumentation:

const instrumentation = new TypeSafeInstrumentation();
instrumentation.setTracerProvider(customTracerProvider);

Compatibility

@arizeai/openinference-instrumentation-typesafe is compatible with the following versions of the @typesafe-ai/sdk package:

TypeSafe SDK Version	OpenInference Instrumentation Version
>=0.6.0	^0.1.0

License

Apache-2.0

This site is open source. Improve this page.