@beacon-ai/vercel

Track AI crawler activity on your Vercel-hosted site. Drop-in Next.js middleware from Beacon.

See which AI providers are crawling your site, how often, and which pages they visit. Beacon identifies 130+ AI agents and classifies them by purpose (training, search, user-initiated).

Install

npm install @beacon-ai/vercel

Setup

Add to your middleware.ts:

import { beaconLogger } from '@beacon-ai/vercel'

export default beaconLogger({
  apiKey: process.env.BEACON_API_KEY!,
  brandId: process.env.BEACON_BRAND_ID!,
})

Or wrap your existing middleware:

import { withBeacon } from '@beacon-ai/vercel'

const myMiddleware = clerkMiddleware(async (auth, request) => {
  // your auth, redirects, etc.
})

export default withBeacon(myMiddleware, {
  apiKey: process.env.BEACON_API_KEY!,
  brandId: process.env.BEACON_BRAND_ID!,
})

Set the environment variables in your Vercel dashboard:

| Variable | Description | |----------|-------------| | BEACON_API_KEY | Your Beacon API key (from your Beacon dashboard) | | BEACON_BRAND_ID | Your brand ID (UUID, from your Beacon dashboard) |

What it captures

Only request metadata. Never request or response bodies.

• Timestamp
• Hostname + path
• HTTP method
• User-Agent
• IP address
• Referer

What it does NOT capture

• Request bodies
• Response bodies
• Cookies
• Auth headers
• Query parameter values (path only)

Performance

Zero impact on response latency. Logging is fire-and-forget with batched async delivery. If the Beacon endpoint is unreachable, logs are silently dropped.

Configuration

beaconLogger({
  apiKey: '...',           // required
  brandId: '...',          // required — your brand ID (UUID)
  batchSize: 50,           // entries to buffer before sending (default: 50)
  flushInterval: 5000,     // ms before flushing buffer (default: 5000)
})

License

MIT