Checkpoint for AI agents · Database v2026-10-18

AI agents stop at your gate.

AgentBlock builds a one-click blocking kit for your site — a Cloudflare Worker, a WordPress plugin, and a JS snippet — from a dated, vendor-documented database of known AI agents. Yelp and the banks blocked them. Now your site can too, in about two minutes.

Build my block kit — free Browse the agent database

No credit card · No code changes to your pages · Free tier included

INBOUND CHECK — DEFAULT POLICY

GPTBot ·············· 403 BLOCK training · OpenAI

ClaudeBot ··········· 403 BLOCK training · Anthropic

facebookexternalhit · ALLOW link previews keep working

unknown-agent/2.0 ··· UNKNOWN never guessed, never silently passed as known

21Verified agents

20Rules blocked by default

25Database entries

3Watchlist — never guessed

3Deploy targets

How it works

Pick a policy. Download the kit. Done.

Step 01

Pick your policy

Allow or block per known AI agent — Meta, OpenAI, Anthropic, Google, Perplexity and more — plus your own allowlist for monitors and partners.

Step 02

Download your kit

A Cloudflare Worker template, a WordPress plugin skeleton, or a plain-JS snippet, generated from your exact policy. Paste, deploy, done.

Step 03

Stay current

The agent database is versioned and dated (v2026-10-18); new agents are added only as vendors document them — never from rumor.

Why it holds up

Built on evidence, not guesses

Known agents only

Every blocked token is vendor-documented. Agents with no published user-agent — Meta's Muse, xAI's Grok, ChatGPT Atlas — are listed as watch and never guessed.

Allowlist first

Your uptime monitors, partners, and link-preview fetchers always win. facebookexternalhit defaults to allow — blocking it breaks link previews.

Three deploy targets

Edge (Cloudflare Worker), server (WordPress plugin), or client (JS snippet deterrent). Team adds an AgentGate agent-policy.json export that AgentGate consumes directly.

Dated and versioned

Every entry carries its evidence label, source, and last-verified date on the database page. Unknown agents classify as UNKNOWN — a first-class answer.

Live check · Team API

Try it: classify a user-agent

Paste any user-agent string — the verdict is BLOCK, ALLOW, or UNKNOWN. Unknown is a first-class answer: we don't guess.

Pricing signal

Free to start. $29 when you need the full kit.

Pro $29/mo unlocks the WordPress plugin and JS snippet. Team $79/mo adds the AgentGate policy export and the classify API. See full pricing

Build my block kit — free Compare plans

Honest scope: HONEST SCOPE: this policy blocks KNOWN agent user-agents and behaviors listed below. Sophisticated spoofing can bypass any user-agent block (e.g. an agent sending a stock browser UA, or rotating residential IPs). This is a control, not a guarantee.