Telemetry

OpenTelemetry integration for AI agent observability. Trace requests, measure latency, and export telemetry to your preferred backend.

Initialization

TypeScript

import { initTelemetry } from '@sekuire/sdk';

await initTelemetry({
  serviceName: 'my-agent',
  exporterUrl: 'https://otel-collector.example.com:4318',
  // or use Sekuire's built-in collector
  useSekuireExporter: true
});

Python

from sekuire_sdk.telemetry import init_telemetry

await init_telemetry(
    service_name="my-agent",
    exporter_url="https://otel-collector.example.com:4318",
    # or use Sekuire's built-in collector
    use_sekuire_exporter=True
)

Rust

use sekuire_sdk::telemetry::init_telemetry;

init_telemetry(TelemetryConfig {
    service_name: "my-agent".into(),
    exporter_url: Some("https://otel-collector.example.com:4318".into()),
    use_sekuire_exporter: true,
    ..Default::default()
})?;

---

Tracing

The SDK automatically creates spans for agent operations. You can also create custom spans:

TypeScript

import { getTracer } from '@sekuire/sdk';

const tracer = getTracer('my-agent');

// Create a span for custom operations
await tracer.startActiveSpan('process_data', async (span) => {
  span.setAttribute('data.size', 1024);

  // Your logic here
  const result = await processData();

  span.setAttribute('result.status', 'success');
  span.end();

  return result;
});

Python

from sekuire_sdk.telemetry import get_tracer

tracer = get_tracer("my-agent")

# Create a span for custom operations
with tracer.start_as_current_span("process_data") as span:
    span.set_attribute("data.size", 1024)

    # Your logic here
    result = await process_data()

    span.set_attribute("result.status", "success")

Rust

use sekuire_sdk::telemetry::get_tracer;

let tracer = get_tracer("my-agent");

tracer.in_span("process_data", |cx| async move {
    cx.span().set_attribute("data.size", 1024);

    // Your logic here
    let result = process_data().await?;

    cx.span().set_attribute("result.status", "success");
    Ok(result)
}).await?;

---

SekuireSpanExporter

Export telemetry directly to Sekuire for integrated dashboard visibility:

TypeScript

import { SekuireSpanExporter } from '@sekuire/sdk';

const exporter = new SekuireSpanExporter({
  backendUrl: 'https://api.sekuire.ai',
  agentId: '7f8a9b3c...',
  batchSize: 100,
  flushInterval: 5000
});

// Integrates with your OTEL setup
provider.addSpanProcessor(
  new BatchSpanProcessor(exporter)
);

Python

from sekuire_sdk.telemetry import SekuireSpanExporter

exporter = SekuireSpanExporter(
    backend_url="https://api.sekuire.ai",
    agent_id="7f8a9b3c...",
    batch_size=100,
    flush_interval=5
)

# Integrates with your OTEL setup
provider.add_span_processor(
    BatchSpanProcessor(exporter)
)

Rust

use sekuire_sdk::telemetry::SekuireSpanExporter;

let exporter = SekuireSpanExporter::new(
    "https://api.sekuire.ai",
    "7f8a9b3c...",
)?
.batch_size(100)
.flush_interval(Duration::from_secs(5));

// Integrates with your OTEL setup

tip

When using useSekuireExporter: true, telemetry data appears in the Sekuire dashboard alongside beacon heartbeats and metrics.

Automatic Instrumentation

The SDK automatically instruments:

• LLM API calls (provider, model, tokens, latency)
• Tool executions (tool name, duration, success/failure)
• Memory operations (store, retrieve, clear)
• Beacon heartbeats
• Policy enforcement checks

Next Steps

• sekuire metrics - View metrics from CLI
• Beacon SDK - Report custom metrics