@visitedby/edge

Server-side AI crawler tracking for visitedby.ai.

AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended — fetch raw HTML. They never execute the client snippet and rarely load images, so client-side analytics is structurally blind to them. This helper runs where your HTML is served and forwards only bot/agent hits to visitedby; human visitors stay with the client snippet, so nothing double-counts.

• Zero dependencies, ~1 KB, Web-standard Request
• Fire-and-forget: never throws, never delays a response (3s hard timeout)
• No new data category: just path + user-agent, same as any server log

Next.js

// middleware.ts
import { NextResponse } from 'next/server'
import { trackAgents } from '@visitedby/edge'

export function middleware(request: Request, event: { waitUntil(p: Promise<unknown>): void }) {
  event.waitUntil(trackAgents(request, { key: process.env.VISITEDBY_SECRET }))
  return NextResponse.next()
}

// pages only — skip assets and API routes
export const config = { matcher: ['/((?!_next|api|.*\\..*).*)'] }

Cloudflare Workers / anything with fetch

import { trackAgents } from '@visitedby/edge'

export default {
  async fetch(request, env, ctx) {
    ctx.waitUntil(trackAgents(request, { key: process.env.VISITEDBY_SECRET }))
    return handle(request)
  },
}

Express / Fastify / Koa / plain Node (18+)

import { trackAgentsNode } from '@visitedby/edge'

app.use((req, _res, next) => {
  void trackAgentsNode(req, { key: process.env.VISITEDBY_SECRET }) // fire-and-forget
  next()
})

SvelteKit (src/hooks.server.ts)

import { trackAgents } from '@visitedby/edge'

export async function handle({ event, resolve }) {
  void trackAgents(event.request, { key: process.env.VISITEDBY_SECRET })
  return resolve(event)
}

Astro (src/middleware.ts)

import { trackAgents } from '@visitedby/edge'

export function onRequest({ request }, next) {
  void trackAgents(request, { key: process.env.VISITEDBY_SECRET })
  return next()
}

Not on a JS runtime? There's a WordPress plugin and a paste-able Cloudflare Worker that works in front of any origin.

Generate your secret ingest key (vbs_…) in the dashboard under Settings → General → Tracking and keep it in an env var — it must never ship to browsers (the public vb_ site key stays in your page HTML for the snippet; server-side forwards using it are rejected once a secret exists). Hits appear on the Traffic page (AI agents & crawlers table) within seconds, each verified against the operator's published IP ranges.