agentwatcher-sdk

One line of code gives you full visibility into every AI API call. Track costs, monitor latency, and catch errors — powered by AgentWatcher.

Install

npm install agentwatcher-sdk

Quick Start

import { AgentWatch } from "agentwatcher-sdk";
import OpenAI from "openai";
import Anthropic from "@anthropic-ai/sdk";

const aw = new AgentWatch({ apiKey: "aw_your_key_here" });

// Wrap OpenAI
const openai = aw.wrapOpenAI(new OpenAI());

// Wrap Anthropic
const anthropic = aw.wrapAnthropic(new Anthropic());

// That's it — all API calls are tracked automatically
const response = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Hello!" }],
});

What Gets Tracked

Every AI API call automatically captures:

• Model — which model was used
• Provider — openai or anthropic
• Token usage — input and output tokens
• Cost — calculated in USD based on current pricing
• Latency — end-to-end response time in ms
• Status — success or error
• Error message — if the call failed

Manual Tracking

aw.track({
  model: "gpt-4o",
  provider: "openai",
  input_tokens: 100,
  output_tokens: 50,
  cost_usd: 0.00175,
  latency_ms: 820,
  status: "success",
});

Configuration

const aw = new AgentWatch({
  apiKey: "aw_your_key_here",   // Required — get yours at agentwatcher.dev
  endpoint: "https://...",      // Optional — defaults to agentwatcher.dev
  debug: true,                  // Optional — logs tracking activity to console
});

Get Your API Key

Sign up free at agentwatcher.dev — no credit card required.

License

MIT