Skip to main content
Mastra emits OpenTelemetry traces following the GenAI semantic conventions through its observability layer. The @mastra/otel-exporter package forwards those spans to any OTLP endpoint — point it at AgentMark and agent runs, model calls, and tool calls arrive normalized. Mastra is a TypeScript framework.

​
Setup

1

Install the exporter

npm install @mastra/observability @mastra/otel-exporter \
  @opentelemetry/exporter-trace-otlp-http
2

Point the exporter at AgentMark

Add an Observability instance with the OTLP exporter to your Mastra app. Use your AgentMark API key and app id from project settings.
import { Mastra } from "@mastra/core/mastra";
import { Observability } from "@mastra/observability";
import { OtelExporter } from "@mastra/otel-exporter";

const observability = new Observability({
  configs: {
    otel: {
      serviceName: "my-app",
      exporters: [
        new OtelExporter({
          provider: {
            custom: {
              endpoint: "https://api.agentmark.co/v1/traces",
              protocol: "http/json",
              headers: {
                Authorization: process.env.AGENTMARK_API_KEY!, // raw key, no "Bearer" prefix
                "X-Agentmark-App-Id": process.env.AGENTMARK_APP_ID!,
              },
            },
          },
        }),
      ],
    },
  },
});

export const mastra = new Mastra({
  agents: { myAgent },
  observability,
});
3

Run your agent

Run your Mastra agent as usual. Each agent run, model call, and tool call arrives in AgentMark as a span, grouped into a trace. See Traces and logs.

​
What AgentMark captures

Mastra’s model spans use the OpenTelemetry GenAI conventions — model, token usage, input and output messages, and finish reason — which AgentMark maps onto its normalized trace fields, and token counts feed cost tracking automatically.

​
Next steps

OpenTelemetry

The endpoint, authentication, and GenAI conventions AgentMark reads

Traces and logs

Explore traces once they arrive

​
Have questions?

Reach out any time: