Adapters

@agentpayxyz/adapters is a single npm package with zero runtime dependencies. It ships adapters for the three most common agent frameworks. Pick what you use — you get payment intents, settlement verification, and AgentPassport trust queries as native tool/function calls.

npm install @agentpayxyz/adapters

Set your API key as an environment variable:

AGENTPAY_API_KEY=sk_live_...

OpenAI function callingopenai

Pass the AgentPay tools to any Chat Completions call. The adapter handles tool execution; your loop just processes the responses.

Register the tools

agent.ts

import OpenAI from 'openai';
import { openaiTools, executeOpenAITool } from '@agentpayxyz/adapters/openai';

const client = new OpenAI();
const agentpayKey = process.env.AGENTPAY_API_KEY!;

const response = await client.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Book the data cleaning job for $2.50' }],
  tools: openaiTools,           // createIntent, verifyPayment, getPassport
  tool_choice: 'auto',
});

for (const choice of response.choices) {
  const msg = choice.message;
  if (msg.tool_calls) {
    for (const call of msg.tool_calls) {
      const result = await executeOpenAITool(call, agentpayKey);
      console.log(result);
      // { intentId: 'intent_01J...', status: 'pending', depositAddress: '7vfC...' }
    }
  }
}

Available tools

Function	Description	Key params
`createIntent`	Create a USDC payment intent	`amount, agentId, metadata?`
`verifyPayment`	Poll until intent verified or expired	`intentId, timeoutMs?`
`getPassport`	Read an agent's trust passport	`agentId`

LangChainlangchain

Structured tools compatible with any LangChain agent executor. Works with LangGraph nodes too.

agent.ts

import { ChatOpenAI } from '@langchain/openai';
import { AgentExecutor, createOpenAIFunctionsAgent } from 'langchain/agents';
import { agentpayLangChainTools } from '@agentpayxyz/adapters/langchain';

const tools = agentpayLangChainTools(process.env.AGENTPAY_API_KEY!);

const llm = new ChatOpenAI({ model: 'gpt-4o', temperature: 0 });
const agent = await createOpenAIFunctionsAgent({ llm, tools, prompt });
const executor = new AgentExecutor({ agent, tools });

const result = await executor.invoke({
  input: 'Create a payment intent for $5 to agent agt_01J...',
});
console.log(result.output);

LangGraph node

graph.ts

import { agentpayLangChainTools } from '@agentpayxyz/adapters/langchain';
import { ToolNode } from '@langchain/langgraph/prebuilt';

const tools = agentpayLangChainTools(process.env.AGENTPAY_API_KEY!);
const toolNode = new ToolNode(tools);

// Add toolNode to your StateGraph as a node

Vercel AI SDKai

Works with generateText, streamText, and the RSC streamUI helper.

app/api/chat/route.ts

import { openai } from '@ai-sdk/openai';
import { streamText } from 'ai';
import { agentpayVercelTools } from '@agentpayxyz/adapters/vercel';

const tools = agentpayVercelTools(process.env.AGENTPAY_API_KEY!);

export async function POST(req: Request) {
  const { messages } = await req.json();

  const result = streamText({
    model: openai('gpt-4o'),
    messages,
    tools,
    maxSteps: 5,
  });

  return result.toDataStreamResponse();
}

All adapters, one package

Every adapter in @agentpayxyz/adapters calls the same AgentPay API under the hood. Switch frameworks without changing your integration logic. Full API reference at api.agentpay.so.