SDK Integration

Ship production traffic to a Monitor from your app.

The PeerLM SDK samples production requests and ships them to a Monitor asynchronously. It never throws into your request path.

Install

npm install @peerlm/sdk
# or
pip install peerlm

TypeScript

import { PeerLMClient } from "@peerlm/sdk";

const peerlm = new PeerLMClient({
  apiKey: process.env.PEERLM_API_KEY!,
  monitorId: "your-monitor-id",
  sampleRate: 0.05,
});

peerlm.track({
  prompt: userMessage,
  systemPrompt,
  response: modelOutput,
  model: "gpt-5.2",
  tokensIn,
  tokensOut,
  latencyMs,
});

Tips

Prefer including response so Runs can use control parity.
Create the Monitor first, paste its ID into the SDK, then wait for events on the Monitor page.