Vercel AI SDK - TraceRoot

Trace Vercel AI SDK agents in TraceRoot. Unlike LangChain, OpenAI, or Anthropic, no instrumentModules configuration is needed — the Vercel AI SDK emits OpenTelemetry spans natively when experimental_telemetry is enabled, and TraceRoot enriches them via the OpenInference span processor automatically.

Setup

npm install @traceroot-ai/traceroot ai @ai-sdk/openai zod

Initialize TraceRoot at your entry point — no instrumentModules config:

import { TraceRoot } from '@traceroot-ai/traceroot';

// No instrumentModules — Vercel AI SDK telemetry is handled automatically.
TraceRoot.initialize();

Usage

import { generateText, tool } from 'ai';
import { openai } from '@ai-sdk/openai';
import { z } from 'zod';
import { observe } from '@traceroot-ai/traceroot';

const result = await observe({ name: 'my_agent', type: 'agent' }, async () => {
  return await generateText({
    model: openai('gpt-4o-mini'),
    maxSteps: 5,
    prompt: 'What is the weather in Tokyo?',
    // This single line activates Vercel AI SDK's built-in OpenTelemetry spans.
    // TraceRoot enriches them automatically.
    experimental_telemetry: { isEnabled: true },
    tools: {
      getWeather: tool({
        description: 'Get current weather for a city',
        parameters: z.object({ city: z.string() }),
        execute: async ({ city }) => ({ city, temp: 72, condition: 'sunny' }),
      }),
    },
  });
});

console.log(result.text);

What Gets Captured

Attribute	Description
Model	The model passed to `generateText` / `streamText` / `generateObject`
Messages	Input prompt and message history
Response	Generated text output
Tool calls	Each tool invocation with input and output
Tokens	Input and output token counts
Cost	Calculated from token usage and model pricing
Latency	Request duration per span
Multi-step iterations	Each `maxSteps` iteration captured as a child span

Run the example

Clone the repo and run a complete agent end-to-end.

TypeScript

Run the TypeScript example

⌘I