Cloudflare Adds Personal Agent Analytics and Per-Agent Access Policies, but Account-Level Views Are Early Access and the Post Does Not List Which Agents Are Supported

On November 4, 2026, the Cloudflare blog described two sets of new features that help site owners see and control "personal agents". A personal agent is software that turns a person's instructions into requests and actions on websites, such as comparing products, finding support articles or completing a purchase. To a site, all of these look like automated traffic. The post argues that blocking agents entirely can cost a site customers, while not knowing which agent does what makes it hard to tell whether the traffic helps or creates risk. This is Cloudflare's own product post and gives no data on results.

The first set is analytics. In Security Analytics, the Bot Analysis chart now shows Personal Assistants as a new category next to Verified Bot, Likely Human, Likely Automated and Automated, so that traffic can be viewed over time on its own. In Top statistics, a new Personal Assistants widget lists which agents visit and how many requests each sends. Filtering on one agent updates every other widget to show only its traffic: the paths it requests, the status codes it receives and the networks it comes from. That shows whether an agent is logging in, reading product pages, reaching checkout, or hitting errors on pages it cannot use. The post says Cloudflare maintains the largest verified bot directory in the world and brings that into traffic details, but it states no accuracy figure for identification and does not list which agents are supported.

For sites that protect user accounts, the post also describes views in Account Abuse Protection and says plainly that this product is currently in early access. A population overview shows which personal agents are active across users and how many accounts use them. Each login and sign-up event shows whether a personal agent made the request and which one, and fraud teams can filter accounts by AI assistant events or by number of AI assistants. A user's account overview lists the agents acting for that account. If an analyst decides an attacker is using an agent on an account, they can write a WAF rule on the account's Hashed User ID to mitigate it, without blocking the agent for everyone else who relies on it. These views matter only to customers who have that early-access product.

The second set is access policy. Cloudflare introduced AI bot policies this summer so that sites can set Search, Training and Agent behaviors separately. The supporting post of July 1 says that from September 15, 2026, new domains onboarding to Cloudflare block Training and Agent bots by default on pages that display ads, while Search stays allowed. Now the existing Agents policy is extended with granular options for supported personal agents, and the section is renamed AI Crawler and Agent access policies. In Security Settings a site sets an overall Agents policy, then picks Allow (do not block), Block on pages with ads, or Block for each supported personal agent. For example, the overall policy can be Block on pages with ads while one agent a site wants to welcome is set to Allow. Existing Search and Training policies stay as they are. The same choices are offered when a new zone is added, where Cloudflare asks whether the site monetizes pages with ads and tailors its recommendation to the answer.

The July 1 supporting post adds two details that bear on these policies. First, from September 15, multi-purpose crawlers that combine Search with Training are allowed or blocked according to all of their behaviors under the most restrictive applicable rule, so crawlers such as Googlebot, Applebot and BingBot are blocked by customers who block Training. Second, the same post introduced BotBase, a searchable database of all known bots, including verified bots and agents, for Enterprise Bot Management customers, which at that point offered visibility only, with controls promised later in the year. The November 4 post does not say whether the new personal agent views draw on BotBase or how supported agents relate to that directory, and readers should not assume it.

The post closes by saying that identity is only part of the story. A site may recognize the agent but lack what it would normally know about the user: whether they are a customer, what they authorized the agent to do, whether they have a subscription or which protections should apply. Cloudflare's next goal is privacy-aware ways to carry user context through an agent, to help assess reputation, and to establish accountability when something goes wrong. These are plans, and the post gives no timeline.

Readers should separate what has shipped from what is still planned. Shipped are the new analytics category and widget and the three-way policy for each supported agent. Early access covers the account-level views. Planned are user context, reputation and accountability. The post does not name the supported agents or the plans that include these features, so check your own Cloudflare dashboard before relying on them.