Backends

autotel-backends provides vendor-specific configuration helpers that simplify init() for popular observability platforms.

If you want to send telemetry to more than one OTLP backend at once, use Autotel’s destinations config in addition to these helpers. Keep endpoint for one destination; use destinations for explicit fan-out.

Installation

npm install autotel autotel-backends

Honeycomb

import { init } from 'autotel';
import { createHoneycombConfig } from 'autotel-backends/honeycomb';

init(
  createHoneycombConfig({
    apiKey: process.env.HONEYCOMB_API_KEY!,
    service: 'my-app',
  }),
);

Datadog

For the full Datadog integration guide – including architecture choices, Agent vs. direct ingestion setup, migration from pino-datadog-transport, deployment patterns, troubleshooting, and advanced configuration – see the dedicated Datadog Integration page.

import { init } from 'autotel';
import { createDatadogConfig } from 'autotel-backends/datadog';

// Direct cloud ingestion (serverless, edge)
init(
  createDatadogConfig({
    apiKey: process.env.DATADOG_API_KEY!,
    service: 'my-lambda',
  }),
);

// Or local Datadog Agent (Kubernetes, long-running services)
init(
  createDatadogConfig({
    service: 'my-api',
    useAgent: true,
  }),
);

Logfire

Logfire requires regional OTLP/HTTP protobuf ingestion. Its write token is not the read token used by autotel investigate --backend logfire.

import { init } from 'autotel';
import { createLogfireConfig } from 'autotel-backends/logfire';

init(
  createLogfireConfig({
    writeToken: process.env.LOGFIRE_WRITE_TOKEN!,
    service: 'my-agent',
    region: 'eu', // required: 'eu' | 'us'
  }),
);

The preset selects http/protobuf, the regional ingest endpoint, and Logfire’s bare-token Authorization format. Use endpoint only for self-hosted Logfire.

Langfuse

[!TIP] This preset covers the connection. For the fields Langfuse keeps in its own columns (trace name, tags, prompt links, time to first token) and for evaluation results as scores, see Langfuse.

import { init } from 'autotel';
import { createLangfuseConfig } from 'autotel-backends/langfuse';

init(
  createLangfuseConfig({
    publicKey: process.env.LANGFUSE_PUBLIC_KEY!,
    secretKey: process.env.LANGFUSE_SECRET_KEY!,
    service: 'my-agent',
    region: 'eu', // 'eu' (default) | 'us'
  }),
);

The preset builds Basic authentication, uses Langfuse’s /api/public/otel receiver, and opts into ingestion v4. Set baseUrl for a self-hosted instance.

PostHog

import { init } from 'autotel';
import { createPostHogConfig } from 'autotel-backends/posthog';

init(
  createPostHogConfig({
    projectToken: process.env.POSTHOG_PROJECT_TOKEN!,
    service: 'my-app',
    region: 'eu', // 'us' (default) | 'eu'
  }),
);

The preset sends traces, logs, and metrics as OTLP/HTTP protobuf, preserves PostHog’s /i path prefix, and uses Bearer authentication. This is OTLP ingest. The session and replay join lives on the PostHog page.

Available Backends

Backend	Import	Transport selected
Honeycomb	autotel-backends/honeycomb	gRPC
Datadog	autotel-backends/datadog	Vendor/mode-specific
Google Cloud	autotel-backends/google-cloud	Vendor-specific
Grafana	autotel-backends/grafana	OTLP/HTTP
Logfire	autotel-backends/logfire	OTLP/HTTP protobuf
Langfuse	autotel-backends/langfuse	OTLP/HTTP JSON
PostHog	autotel-backends/posthog	OTLP/HTTP protobuf

Multi-backend fan-out

Use destinations when you want one app to ship to multiple OTLP backends without manually constructing exporters:

import { init } from 'autotel';

init({
  service: 'my-app',
  logs: true,
  destinations: [
    {
      endpoint: 'https://otlp-gateway-prod.grafana.net/otlp',
      headers: 'Authorization=Basic ...',
    },
    {
      endpoint: 'https://api.honeycomb.io',
      headers: { 'x-honeycomb-team': process.env.HONEYCOMB_API_KEY! },
      signals: ['traces'],
    },
  ],
});

Use the backend helpers when you want vendor-specific defaults for one destination. Use destinations when you want one service to fan out to several OTLP backends.

Examples

• example-datadog: Datadog with createDatadogConfig(), direct cloud ingestion and local agent modes.
• example-grafana: Grafana Cloud (Tempo + Mimir + Loki) via OTLP with canonical log lines.

Backends vs Plugins

Package	Purpose
autotel-backends	Configure where telemetry goes (outputs)
autotel-plugins	Instrument libraries to create telemetry (inputs)