Track GPTBot and ClaudeBot on Next.js and Vercel

Fallback: add it by hand

2. Install the package

Prefer to wire it yourself? Install the @foglift/sensor runtime the CLI would have added.

npm install @foglift/sensor@1.0.0

3. Add the generated boundary file

Open AI Traffic while signed in to copy this file with your workspace identifier filled in. The matcher excludes static Next.js assets while keeping page requests observable. On Next.js 16+, save it as proxy.ts and rename the export to proxy.

// middleware.ts
import { senseAITraffic } from "@foglift/sensor/nextjs";

export const middleware = senseAITraffic({
  siteToken: "YOUR_WORKSPACE_ID",
});

export const config = {
  matcher: ["/((?!_next/static|_next/image|favicon.ico).*)"],
};

4. Preserve existing middleware

If the project already has middleware, compose it through the next option. Do not replace existing authentication, redirects, headers, or routing behavior.

// middleware.ts
import { senseAITraffic } from "@foglift/sensor/nextjs";
import { existingMiddleware } from "./src/middleware-existing";

export const middleware = senseAITraffic({
  siteToken: "YOUR_WORKSPACE_ID",
  next: existingMiddleware,
});

Keep the Next.js request event parameter available. The Sensor retains its report for the request lifetime with event.waitUntil, without blocking the response.

5. Deploy to Vercel and verify

1. Run the project tests and production build before deploying.
2. Deploy the middleware or proxy file with the Next.js application on Vercel, then open AI Traffic.
3. Choose Verify install. Foglift sends a unique workspace-bound request and confirms when the middleware reports it.

A crawler request proves the agent reached the page. It does not prove indexing, citation, recommendation, or model training.