@paid-ai/instrument-repo

Automatically instrument your repository with the Paid platform for AI agent tracing and analytics. 5-phase setup process that detects your codebase, sets up dependencies, and transforms your AI SDK code.

Usage

npx @paid-ai/instrument-repo

Generate your free API key at https://app.paid.ai/.

What it does

The tool follows a structured 5-phase process:

1. Analysis Phase - Auto-detects language, framework, package manager, and existing setup
2. Planning Phase - Asks for missing information with detected values as defaults
3. Setup Phase - Installs packages, creates .env files, and sets up client utilities
4. Detection Phase - Finds AI SDK method calls that need instrumentation
5. Transformation Phase - Uses LLM to transform your code with Paid tracing

Supported Languague Framework Combinations

Currently supports:

• Node.js + Vercel AI SDK projects

What gets created

• utils/client.ts (or similar) - Paid client initialization
• .env file with your Paid API key
• Transforms AI SDK calls (streamText, generateObject, generateText) to use Paid tracing

Requirements

• Paid API Key - For platform access and tracing
• OpenAI API Key - For LLM code transformations
• Customer ID - Used in tracing calls to identify your users

Example Transformation

Before:

import { streamText } from 'ai';

const result = await streamText({
  model: openai('gpt-4'),
  prompt: 'Hello'
});

After:

import { paidStreamText } from '@paid-ai/paid-node/vercel';
import { getClient } from '../utils/client';

const client = await getClient();
const result = await client.trace('customer-123', async () => {
  return await paidStreamText({
    model: openai('gpt-4'),
    prompt: 'Hello'
  });
});