Use 402.report with the Vercel AI SDK
Route your Vercel AI SDK agent's HTTP through 402.report and expose its x402 spend as MCP tools — set one API key.
The Vercel AI SDK speaks MCP (@ai-sdk/mcp /
experimental_createMCPClient) and accepts a custom fetch. 402.report plugs into both: expose
your spend as tools, and observe the payments your agent makes.
1. Install
npm install @402report/sdk ai
Set your key as REPORT402_API_KEY (create one in the dashboard).
2. Add the spend tools
import { experimental_createMCPClient as createMCPClient, generateText } from "ai";
import { openai } from "@ai-sdk/openai";
import { aiSdkMcpTransport } from "@402report/sdk";
const mcp = await createMCPClient({
transport: aiSdkMcpTransport({ apiKey: process.env.REPORT402_API_KEY! }),
});
const { text } = await generateText({
model: openai("gpt-4o"),
tools: await mcp.tools(),
prompt: "How much have my agents paid this week, and did any payments fail?",
});
await mcp.close();
3. Observe the payments your agent makes
Route model/tool HTTP through the proxy with a drop-in fetch so every x402 challenge and
payment is recorded:
import { createProxyFetch } from "@402report/sdk";
import { createOpenAI } from "@ai-sdk/openai";
const openai = createOpenAI({
apiKey: process.env.OPENAI_API_KEY!,
fetch: createProxyFetch({ apiKey: process.env.REPORT402_API_KEY! }),
});
402.report is non-custodial — it observes x402 activity, it never holds or settles funds.