@quetra/sdk

TypeScript client library for AI agent spending governance. Works with any agent framework — LangChain, Claude Agent SDK, Vercel AI SDK, or custom implementations.

Installation

npm install @quetra/sdk
# or
pnpm add @quetra/sdk

Quick Setup

import { QuetraClient } from "@quetra/sdk";

const client = new QuetraClient({
  apiKey: process.env.QUETRA_API_KEY!,
  agentId: "your-agent-uuid",
  gatewayUrl: "https://gateway.quetra.dev", // default
});

Methods

Method	Description
evaluate(opts)	Evaluate a transaction against mandate rules
fetch(url, opts)	HTTP fetch with automatic x402 payment handling
getActiveMandate()	Get the agent's active mandate
getBudgetStatus()	Check remaining budget and limits
canSpend(opts)	Check if a spend would be approved, with rejection reasons
getTransactions(opts)	List recent transactions
acpCheckout(opts)	Initiate Stripe ACP merchant checkout
stripeCharge(opts)	Evaluate and create Stripe PaymentIntent

x402 Payment Flow

client.fetch() automatically handles the x402 payment protocol:

// 1. Initial request to paid API
// 2. Receives 402 Payment Required with PAYMENT-REQUIRED header
// 3. Parses payment requirements (amount, vendor, category)
// 4. Calls gateway /x402/pay to evaluate + sign payment
// 5. Retries original request with PAYMENT-SIGNATURE header
// 6. Returns the API response

const data = await client.fetch("https://paid-api.example.com/research");

Environment Variables

QUETRA_API_KEY=sk_your_api_key_here
QUETRA_AGENT_ID=your-agent-uuid
QUETRA_GATEWAY_URL=https://gateway.quetra.dev  # optional, default