Ship device intelligence with agents today

Connect your AI tools to Fingerprint, then configure and add device intelligence to your app straight from your editor or terminal.

Or open in Claude Code, Codex, Cursor

AI agents

Use Fingerprint with AI

Have your agent add Fingerprint to your app, new or existing, without leaving your editor. Works with the tools your team already uses.

AI docs

llms.txt

SKILLS.md

Give your AI the skills
to build with Fingerprint

Full, up-to-date documentation without copy-pasting. Enable AI to detect your stack, install the correct SDK, wire up identification, and add the server-side verification that runs Fingerprint securely.

Get API keys

Cursor

Install the Fingerprint plugin from the Cursor Directory.
View in directory

Claude Code

Add the plugin repository, then install in your project.

1 claude plugin marketplace add fingerprintjs/skills

2 claude plugin install fingerprint@fingerprint

Codex

Add the plugin repository, then install in your project.

1 codex plugin marketplace add fingerprintjs/skills

2 codex plugin add fingerprint@fingerprint

CLI

Add Fingerprint by CLI
using one command

The CLI detects your stack, provisions API keys for your workspace, and writes the integration directly in your app: client-side identification, server-side verification, and a protected action wired end-to-end.

GitHub

~/Dev/dashboard

fingerprint

integrate

whoami

dusan@MacBook-Pro dashboard % ╭─                                                                                      ─╮   ███████╗██╗███╗   ██╗ ██████╗ ███████╗██████╗ ██████╗ ██████╗ ██╗███╗   ██╗████████╗      ██╔════╝██║████╗  ██║██╔════╝ ██╔════╝██╔══██╗██╔══██╗██╔══██╗██║████╗  ██║╚══██╔══╝      █████╗  ██║██╔██╗ ██║██║  ███╗█████╗  ██████╔╝██████╔╝██████╔╝██║██╔██╗ ██║   ██║         ██╔══╝  ██║██║╚██╗██║██║   ██║██╔══╝  ██╔══██╗██╔═══╝ ██╔══██╗██║██║╚██╗██║   ██║         ██║     ██║██║ ╚████║╚██████╔╝███████╗██║  ██║██║     ██║  ██║██║██║ ╚████║   ██║         ╚═╝     ╚═╝╚═╝  ╚═══╝ ╚═════╝ ╚══════╝╚═╝  ╚═╝╚═╝     ╚═╝  ╚═╝╚═╝╚═╝  ╚═══╝   ╚═╝      ╰─                                                                               v0.1.0 ─╯ Welcome to the Fingerprint CLI. Fingerprint identifies returning browsers and devices with a stable visitor ID. Smart Signals add device intelligence to help detect bots and suspicious activity. This CLI sets up Fingerprint end to end. It detects your stack, provisions API keys, and writes the integration into your code. Sign in or create an account to get started. Commands fingerprint           this guided setup, start to finish fingerprint integrate add Fingerprint to the repo in this directory fingerprint keys      print a public or secret API key fingerprint whoami    show the signed-in workspace fingerprint --help    every command and flag✔ Do you already have a Fingerprint account? Yes — log in◇ Sign in│ Opening your browser...│ │ Waiting for you to finish in the browser. Please return here when you’re done...✔ Signed in workspace sub_fp39EpE2neBgjs│ integrate│ ◇ Analyzing project│ Repository  ~/Dev/fingerprint-sample-stacks/stacks/nextjs│ Layout      single app│ │ Apps found│   nextjs JavaScript · npm · next│ │ Frontend    next (.)│ Backend     not found│ │ Skills      fingerprint-nextjs

MCP server

Use the Fingerprint MCP
for natural language analysis

Point Claude, ChatGPT, Cursor or any MCP client⁰¹ at Fingerprint to query events, detect anomalies, investigate fraud, and manage your workspace through natural language prompts.

MCP docs

Claude

Add to Claude
Used Fingerprint integration, used a tool
Total events
100
Datacenter
80
Residential proxy
15
ISP / direct
5
Bot type
Webdriver63
Headless Chrome18
Puppeteer Stealth5
AdsPower SunBrowser4
Nodriver4
Browser Automation Studio4
Nodriver (unclassified)2
Country of origin
US — Des Moines, IA63
US — Ashburn, VA9
France — Paris/Nantes5
India — Patna4
A few things stand out:
One visitor dominates. Visitor zEAhZ8fx... accounts for 63 of the 100 events — all webdriver, all from a single Microsoft Azure IP in Des Moines. It’s clearly an automated test suite (the URLs match fingerprint.com flows like /playground, /personalization, /paywall, /sms-pumping) hitting your demo environment, not a real attack. Probably internal or partner testing that doesn’t bother suppressing bot detection.
The more interesting signals are the others. The residential proxy events are the ones worth watching:
AdsPower SunBrowser + residential proxies (NetNut, DataImpulse) hitting fingerprint.com from French ISP IPs. Anti-detect browser paired with resi proxy is a classic account fraud pattern.
Browser Automation Studio from India via PROXY.CC on Airtel IPs, landing on fingerprint.com. Someone testing their own tooling against your detection.
Puppeteer Stealth out of Brazil with 711Proxy, coming in via a Google Ads click (the URL has full UTM params). That’s potentially click fraud.
Network type breakdown: 80% datacenter (mostly that Azure cluster), 15% residential proxy, 5% ISP/direct. If you strip the Azure test suite out, the residential proxy ratio jumps to ~75% of the remaining organic bad bot traffic.
ChatGPT

Add to ChatGPT

Searched available tools
I’ll run several searches in parallel across the major suspicious signal categories over the past 30 days.
Used Fingerprint integration
Good data. Let me pull one more targeted query to look at the high-suspect-score non-bot traffic and the AdsPower cluster more specifically.
Search Event History
Good data across all three queries. Here’s what stands out:
Cluster 1: Automated demo scraper (visitor zEAhZ8fx39V8hFn0V7Eb) — highest priority
This is the most active suspicious visitor in the dataset. A single visitor ID generated 200+ events.
Cluster 2: HeadlessChrome burst on your homepage
Multiple new visitor IDs all using HeadlessChrome/139 on Linux, all hitting fingerprint.com/ in rapid sequence from the same source. VPN-flagged. Each generates a fresh visitor ID (not seen before), which combined with the headless UA and near-identical timing is a classic scraping or recon pattern.
Cluster 3: AdsPower SunBrowser — 3+ distinct sessions
Visitor s0naFmwPqB2dB2g6o75R and others are confirmed AdsPower anti-detect browser sessions (bot_info.name: "AdsPower SunBrowser"). They spoof iOS/iPhone UAs while running on what’s clearly automation infrastructure, combine proxy + tampering (anomaly_score: 1.0), and appear across multiple sessions on fingerprint.com/. This is explicitly browser fingerprint evasion tooling — worth flagging as a product-specific signal validation case.
Cluster 4: Automated job application bot (visitor doE6m3aWK0jDIrWbKWwk)
Linux/aarch64, Chrome 120, repeatedly hitting /careers/jobs/apply/ with different gh_jid params across 18 events in 24 hours from multiple IPs. VPN on. Looks like an automated job scraper or applicant bot cycling through open roles.
Cluster 5: Tampering without bad bot label — older Chrome UA pattern
Several events tagged tampering: true with anomaly_score ~0.9994 show Chrome 129 on Windows 10 — a notably stale version that’s ~1.5 years behind current. No bot detection triggered, suspect scores 16-30. These could be headless setups that didn’t trigger bot detection but are showing fingerprint inconsistencies. Also a handful of aarch64 Linux + Chrome 139 events in the same bucket, coming through VPN, referrer Google.
Notable one-off: WebView
One event from a Samsung Galaxy Z Fold5 WebView, suspect score 58, tampering anomaly_score 1.0, VPN on, bot flagged. Could be a fraud probe or someone doing recon via an embedded browser in a crypto app.
The total_hits cap on both bad bot and tampering queries hit 1,000, so this is a dense period of activity. The demo site is clearly taking a lot of automated traffic — probably expected given it’s publicly accessible, but Cluster 1’s volume is notable enough to warrant a closer look.
Cursor

Add to Cursor
Worked for 3m 54s
Thought briefly
I’ll start by auditing the project for its stack and any existing Fingerprint usage.
List project files and manifest cd, ls, cat, git, head
total 0 drwxr-xr-x 2 user staff 64 Sep 28 09:15 . drwx------+ 22 user staff 704 Sep 28 09:15 ..
Explored 1 search
Answer
The workspace folder is empty. How should I proceed?
Scaffold a new Next.js app here (App Router, TypeScript), then add Fingerprint (Recommended)
Thought briefly
I’ll scaffold the app and pull the current Fingerprint Next.js and Smart Signals guides at the same time.
Fetch Fingerprint skill guides echo, curl
yourself. - Region must match the workspace region (`us` | `eu` | `ap`) on every side. - If the user only wants one step, jump straight to that step's skill.
The app is scaffolded.
Thought briefly
Next I’ll install both SDKs. While that runs, I’ll check the Fingerprint MCP for your workspace’s API keys and region.
Thought 18s
fingerprint-provider.tsx
+20
1"use client";
2
3import { FingerprintProvider } from "@fingerprint/react";
4
layout.tsx
+3-1
25>
26<body className="min-h-full flex flex-col">
26<body className="min-h-full flex flex-col antialiased">
27 <FingerprintClientProvider>{children}
layout.tsx
+1
3import "./globals.css";
4import { FingerprintClientProvider } from "./fingerprint-provider";
5
Now the server-side module: it verifies the event and applies a bot and AI policy that depends on how risky the action is.
fingerprint-server.ts
+164
1import "server-only";
2import {
3 FingerprintServerApiClient,
4 Region,
route.ts
+24
1import { verifyEvent } from "@/lib/fingerprint-server";
2
3export async function POST(request: Request) {
4 const { username, eventId } = (await request.json())
Thought briefly
Next, the login form. It identifies the visitor only when the form is submitted and shows what the server decided.
login-form.tsx
+110
1"use client";
2
3import { useState } from "react";
4import { useVisitorData } from "@fingerprint/react";
Thought briefly
I scaffolded a new Next.js 16 app (the folder was empty) and added Fingerprint identification plus server-side bot and AI-agent detection on a demo login form. Type-check, lint, and next build all pass. I haven’t tested it against Fingerprint yet because the API keys aren’t filled in. Auto-review blocked me from reading your keys through the Fingerprint MCP, so I left them out of the chat.
What to do next
Get a public key and a secret key from the dashboard (App settings → API keys). Your workspace has two environments, "Production" and "Staging", so take both keys from the same one.
Paste them into .env.local. It’s already created and gitignored. Set both region variables to your workspace region (us, eu or ap).
Run npm run dev, submit the form, and check that the event appears on the dashboard Events page. If nothing shows up, turn off your ad blocker and try again.

Frequently asked questions

Fingerprint gives AI coding agents the tools and guidance to integrate device intelligence into your app, from installing the right SDK to adding server-side verification. Through the Fingerprint MCP server, AI tools can also query your identification data, investigate suspicious activity, and manage your workspace using natural language.

Integrate Fingerprint with AI today

Onboard your AI coding agent and it handles the rest.

Read the docs

Ship device intelligence with agents today

Use Fingerprint with AI

Claude Code

Codex

Cursor

Devin Desktop

OpenCode

Pi

VS Code

Zed

Give your AI the skills
to build with Fingerprint

Cursor

Claude Code

Codex

Add Fingerprint by CLI
using one command

Use the Fingerprint MCP
for natural language analysis

Claude

Bot type

Country of origin

ChatGPT

Cluster 1: Automated demo scraper (visitor `zEAhZ8fx39V8hFn0V7Eb`) — highest priority

Cluster 2: HeadlessChrome burst on your homepage

Cluster 3: AdsPower SunBrowser — 3+ distinct sessions

Cluster 4: Automated job application bot (visitor `doE6m3aWK0jDIrWbKWwk`)

Cluster 5: Tampering without bad bot label — older Chrome UA pattern

Notable one-off: WebView

Cursor

What to do next

Frequently asked questions

Integrate Fingerprint with AI today

Ship device intelligence with agents today

Use Fingerprint with AI

Claude Code

Codex

Cursor

Devin Desktop

OpenCode

Pi

VS Code

Zed

Give your AI the skillsto build with Fingerprint

Cursor

Claude Code

Codex

Add Fingerprint by CLI using one command

Use the Fingerprint MCP for natural language analysis

Bot type

Country of origin

Cluster 1: Automated demo scraper (visitor zEAhZ8fx39V8hFn0V7Eb) — highest priority

Cluster 2: HeadlessChrome burst on your homepage

Cluster 3: AdsPower SunBrowser — 3+ distinct sessions

Cluster 4: Automated job application bot (visitor doE6m3aWK0jDIrWbKWwk)

Cluster 5: Tampering without bad bot label — older Chrome UA pattern

Notable one-off: WebView

What to do next

Frequently asked questions

01How does Fingerprint work with AI agents?

02Which agents and tools does Fingerprint support?

03Which frameworks can an AI agent set up Fingerprint in?

04Do I need an API key before prompting my agent?

05Is the Fingerprint MCP server free to use?

Integrate Fingerprint with AI today

Give your AI the skills
to build with Fingerprint

Add Fingerprint by CLI
using one command

Use the Fingerprint MCP
for natural language analysis

Cluster 1: Automated demo scraper (visitor `zEAhZ8fx39V8hFn0V7Eb`) — highest priority

Cluster 4: Automated job application bot (visitor `doE6m3aWK0jDIrWbKWwk`)