Trust & security

Cloudflare verified bots and signed agents: what changed in July 2026

Listen · 10 min

Narrated by an AI voice. Code examples are on the page.

Short answer

A Cloudflare verified bot is a bot or agent that Cloudflare has confirmed identifies itself honestly and behaves well. Signed agents used to be a separate category for agents that sign their requests with Web Bot Auth. Since 1 July 2026 they are part of Verified, and Cloudflare marks the difference with a Direct or Intermediary label instead.

Key takeaways

  • Since 1 July 2026, Cloudflare counts signed agents as verified bots.
  • Direct means one operator; Intermediary means many end users drive it.
  • Set Search, Agent and Training actions in Cloudflare's security settings.

What is a Cloudflare verified bot?

A verified bot is a bot or agent that Cloudflare has checked and confirmed. Cloudflare's documentation describes it as one that is "transparent about who it is and what it does". Examples include search engine crawlers, monitoring services and agents driven by users.

Cloudflare sets two bars. The first is honest self-identification: the bot declares who it is through a Web Bot Auth signature, a published IP list with a stable user-agent, or reverse DNS. The second is non-abusive behavior: it obeys robots.txt and crawl directives, keeps request rates reasonable, and has not been seen evading site owner preferences or attacking sites.

Both bars matter. A name in a user-agent header proves nothing on its own, as our verified bot definition explains. Cloudflare's label adds a behavior review on top of the identity check.

Cloudflare can take the label away. Its docs say that if a requirement is breached, "a service will be removed from the global allowlist". One listed example is an AI crawler that ignores the crawl-delay directive in robots.txt.

What were signed agents?

Signed agents were a category Cloudflare introduced on 28 August 2025 for agents that act for many different end users and sign their requests.

Cloudflare's announcement gave three conditions. The agents are "generally directed by an end user instead of a single company or entity". The infrastructure or remote browser they use signs their requests with Web Bot Auth, and Cloudflare checks those signatures. And they follow Cloudflare's signed agent policy.

The first cohort Cloudflare named was ChatGPT agent, Goose from Block, Browserbase and Anchor Browser. Cloudflare's own Browser Rendering product was placed in the same category. The idea was that a crawler like a search bot works for one company, while a cloud browser works for whoever is using it.

What changed on 1 July 2026?

Signed agents stopped being a separate category. Cloudflare's verified bots page now says: "Signed agents are now Verified."

The page goes on: "As of July 1, 2026, the distinction between a Verified bot and a signed agent is expressed by a new metadata field tracked in BotBase: Direct versus Intermediary access, which tracks who can operate the bot."

Label What Cloudflare says it means
Direct Run by "a single, narrow operator". Only that operator can send requests that present as this bot.
Intermediary "An agentic service that a wide range of end users can operate." The operator runs the software, but different end users start each action.

For a site owner, the Intermediary label is the one to think about. Cloudflare calls the issue transitive trust: "you may trust the intermediary operator, but not necessarily every end user driving it." A verified cloud browser tells you which company runs the browser. It does not tell you who is typing the instructions. Cloudflare says it is experimenting with passing on information about the end user in the Forwarded header.

The same update brought in a new way to classify bots by behavior. A bot can have one or more of these: Search, Agent, Training, Transact, Data Collection, Security Testing, SEO, Ads Verification, Social / Link Preview, Feed Fetching, and Monitoring & Operations. Cloudflare says these new behaviors "are not individual fields in WAF custom rules". The older categories, such as AI Crawler and AI Search, still work in rules.

How do you see and control verified bots and AI agents in Cloudflare?

You can see them in three places, and control them with behavior settings, per-crawler choices and custom rules. Which tools you have depends on your plan.

  • AI Crawl Control. Available on all plans. Its Crawlers tab lists the AI crawlers requesting your content, with the operator, category, request counts and robots.txt violations. On the free plan it identifies crawlers by their user-agent strings, so it sees the well-known bots that name themselves. Cloudflare says upgrading your plan enables more thorough detection.
  • BotBase. Cloudflare's directory of all the bots it tracks, inside the dashboard under Security Analytics, Bot analysis, BotBase. It shows each bot's behaviors and lets you filter your own traffic to one bot. It is available to Enterprise Bot Management customers.
  • Cloudflare Radar. Cloudflare says every bot tracked in BotBase is listed publicly in Radar's bots and agents directory, so anyone can look up how a bot is classified.

To control them, start with the behavior settings, then use per-crawler choices or custom rules for exceptions.

AI Crawler and Agent access policies. In the dashboard, go to Security, then Settings, filter by Bot traffic and open AI Crawler and Agent access policies. You choose an action for Search, Agent and Training. The actions are Block, Block on pages with ads, Allow (do not block), and, for Training only, Disallow, which adds a robots.txt instruction. Cloudflare says "Each policy applies to Verified bots with that behavior and matching unverified bots."

For new domains, all three behaviors default to Allow (do not block). Personal Agents inherit the Agent setting, and you can then set individual ones. Cloudflare recommends Allow (do not block) for Personal Agents.

Per-crawler actions. In AI Crawl Control you can allow or block each AI crawler. When you block one, Cloudflare creates or updates a WAF custom rule on your zone to enforce it, which you can extend with exceptions.

Custom rules. Cloudflare's rules language includes fields for this. cf.client.bot indicates a request from a known good bot. cf.verified_bot_category holds the older category name, such as AI Crawler. On Enterprise plans with Bot Management, cf.bot_management.signed_agent marks agents that identify themselves with Web Bot Auth, which Cloudflare now classes as verified and labels Intermediary. An illustrative rule expression that matches verified AI training crawlers:

(cf.verified_bot_category eq "AI Crawler")

Remember that robots.txt rules are voluntary. Cloudflare's own docs say crawlers can ignore them, which is why the edge actions exist. Our robots.txt guide covers which bots to allow there.

How does this relate to Web Bot Auth?

Web Bot Auth is one of the ways a bot proves its identity to Cloudflare. Verified status is that proof plus a behavior review.

Cloudflare accepts two validation methods for verified bots: Web Bot Auth and IP validation. With Web Bot Auth, the operator hosts its public keys at /.well-known/http-message-signatures-directory, registers that directory with Cloudflare, and signs each request. Cloudflare then verifies those signatures. The mechanics are in our Web Bot Auth explainer.

So the three ideas fit together like this. A signature proves which operator sent a request. The Verified label says Cloudflare also reviewed how that operator's bot behaves. The Direct or Intermediary label says whether one company or many end users are behind each request. None of them tells you whether a particular person was allowed to buy something. That is a separate question, covered in our piece on know your agent (KYA).

What should a store owner do now?

  1. Review your three behavior settings. Decide what Search, Agent and Training traffic may do. Some stores want search and shopping agents to reach product pages while treating training crawlers differently.
  2. Look at what is visiting. Open AI Crawl Control and see which crawlers come, how often and whether they follow robots.txt.
  3. Treat Intermediary agents with care at checkout. A verified cloud browser is a known operator, not a known shopper. Keep your normal order and fraud checks.
  4. Do not trust names alone. A request that only claims a well-known name in its user-agent has not proved who sent it. Our guide on telling whether a visitor really is ChatGPT shows how.
  5. Check how your site looks to agents. The free readiness scan shows what AI agents can find today.

How Sill helps

Sill shows you which AI agents and bots visit your site and the pages they read. On Cloudflare and WordPress installs, it verifies signed agent requests (Web Bot Auth) for agents that sign, so you can see which visitors proved who they are and which only claimed to.

Questions people ask

Is a Cloudflare verified bot always safe to allow?

Verified means Cloudflare confirmed who the bot is and that it meets its behavior rules. It does not mean you want it. You still choose what each behavior, such as Training or Agent, may do on your site.

Can I see AI bots on Cloudflare's free plan?

Yes. AI Crawl Control is available on all plans and lists the AI crawlers visiting your site. On the free plan it identifies them by their user-agent strings, so it catches well-known bots that name themselves.

Will blocking AI training crawlers also block shopping agents?

Not by itself. Cloudflare treats Training and Agent as separate behaviors with separate actions. One bot can have more than one behavior, so check how a bot is classified before you rely on a single setting.

How does a bot become a Cloudflare verified bot?

Its operator applies through the BotBase submission form in the Cloudflare dashboard. The bot must be verified by Web Bot Auth or by IP validation, and it must follow Cloudflare's behavior policy.

Sources

  1. Cloudflare docs: Verified bots · checked 16 Nov 2026
  2. Cloudflare docs: Bots (AI bots classification) · checked 16 Nov 2026
  3. Cloudflare docs: AI Crawler and Agent access policies · checked 16 Nov 2026
  4. Cloudflare docs: BotBase · checked 16 Nov 2026
  5. Cloudflare docs: Manage AI crawlers (AI Crawl Control) · checked 16 Nov 2026
  6. Cloudflare docs: AI Crawl Control · checked 16 Nov 2026
  7. Cloudflare docs: Bot Management variables · checked 16 Nov 2026
  8. Cloudflare docs: cf.client.bot field · checked 16 Nov 2026
  9. Cloudflare docs: cf.bot_management.signed_agent field · checked 16 Nov 2026
  10. Cloudflare docs: Web Bot Auth · checked 16 Nov 2026
  11. Cloudflare blog: The age of agents: cryptographically recognizing agent traffic · checked 16 Nov 2026

Written by The Sill Signals Team. Last reviewed 16 Nov 2026.

Share this articleLinkedInXBlueskyEmail

Keep reading

See which AI agents visit your site, and decide what they may do.

Scan your site free