TypeScript SDK Reference

npm install @aicostguard/sdk


# or


yarn add @aicostguard/sdk


# or


pnpm add @aicostguard/sdk

import { CostGuard } from '@aicostguard/sdk';





const guard = new CostGuard({


  apiKey: 'cg_your_key_here',     // Required


  projectId: 'my-project',        // Optional — groups costs by project


  environment: 'production',      // Optional — dev | staging | production


  baseUrl: 'https://api.aicostguard.com', // Optional — self-hosted override


  flushInterval: 5000,            // Optional — batch send interval (ms)


  debug: false,                   // Optional — enable debug logging


});

import OpenAI from 'openai';





const openai = guard.wrap(new OpenAI());

import Anthropic from '@anthropic-ai/sdk';





const anthropic = guard.wrap(new Anthropic());

import { GoogleGenerativeAI } from '@google/generative-ai';





const genai = guard.wrap(new GoogleGenerativeAI(apiKey));

const response = await openai.chat.completions.create({


  model: 'gpt-4o-mini',


  messages: [{ role: 'user', content: 'Hello' }],


}, {


  costGuard: {


    userId: 'user_123',


    feature: 'customer-support',


    tags: { sprint: '24', ticket: 'JIRA-456' },


  },


});

await guard.budgets.create({


  name: 'Production Daily Cap',


  amount: 100,              // $100


  period: 'daily',          // daily | weekly | monthly


  alertThresholds: [50, 80, 100],


  autoStop: true,           // Pause API calls at 100%


  notifyChannels: ['slack', 'email'],


});

const costs = await guard.costs.query({


  from: '2026-03-01',


  to: '2026-03-15',


  groupBy: ['model', 'feature'],


  projectId: 'my-project',


});





console.log(costs);


// [{ model: 'gpt-4o', feature: 'chat', totalCost: 142.50, requests: 23400 }, ...]

try {


  const chat = await openai.chat.completions.create({ ... });


} catch (error) {


  // CostGuard never blocks your API calls


  // If tracking fails, your LLM call still succeeds


  // Errors are logged to guard.on('error', handler)


}