Changelog

Users can now use an OpenAI Compatible endpoint in AI Gateway to easily switch between providers, while keeping the exact same request and response formats. We're launching now with the chat completions endpoint, with the embeddings endpoint coming up next.

To get started, use the OpenAI compatible chat completions endpoint URL with your own account id and gateway id and switch between providers by changing the model and apiKey parameters.

import OpenAI from "openai";
const client = new OpenAI({
	apiKey: "YOUR_PROVIDER_API_KEY", // Provider API key
	baseURL:
		"https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/compat",
});

const response = await client.chat.completions.create({
	model: "google-ai-studio/gemini-2.0-flash",
	messages: [{ role: "user", content: "What is Cloudflare?" }],
});

console.log(response.choices[0].message.content);

Additionally, the OpenAI Compatible endpoint can be combined with our Universal Endpoint to add fallbacks across multiple providers. That means AI Gateway will return every response in the same standardized format, no extra parsing logic required!

Learn more in the OpenAI Compatibility documentation.

Shopify merchants can now onboard to O2O automatically, without needing to contact support or community members.

What's new:

• Automatic enablement – O2O is available for all mutual Cloudflare and Shopify customers.

• Branded record display – Merchants see a Shopify logo in DNS records, complete with helpful tooltips.

  

• Checkout protection – Workers and Snippets are blocked from running on the checkout path to reduce risk and improve security.

For more information, refer to the provider guide.

You can now visualize, explore and modify your Worker’s architecture directly in the Cloudflare dashboard, making it easier to understand how your application connects to Cloudflare resources like D1 databases, Durable Objects, KV namespaces, and more.

With this new view, you can easily:

• Explore existing bindings in a visual, architecture-style diagram
• Add and manage bindings directly from the same interface
• Discover the full range of compute, storage, AI, and media resources you can attach to your Workers application.

To get started, head to the Cloudflare dashboard ↗ and open the Bindings tab of any Workers application.

We're excited to announce the Public Beta launch of User Groups for Cloudflare Dashboard and System for Cross Domain Identity Management (SCIM) User Groups, expanding our RBAC capabilities to simplify user and group management at scale.

We've also visually overhauled the Permission Policies UI to make defining permissions more intuitive.

What's New

User Groups [BETA]: User Groups are a new Cloudflare IAM primitive that enable administrators to create collections of account members that are treated equally from an access control perspective. User Groups can be assigned permission policies, with individual members in the group inheriting all permissions granted to the User Group. User Groups can be created manually or via our APIs.

SCIM User Groups [BETA]: Centralize & simplify your user and group management at scale by syncing memberships directly from your upstream identity provider (like Okta or Entra ID) to the Cloudflare Platform. This ensures Cloudflare stays in sync with your identity provider, letting you apply Permission Policies to those synced groups directly within the Cloudflare Dashboard.

Revamped Permission Policies UI [BETA]: As Cloudflare's services have grown, so has the need for precise, role-based access control. We've given the Permission Policies builder a visual overhaul to make it much easier for administrators to find and define the exact permissions they want for specific principals.

For more info:

• Get started with User Groups
• Explore our SCIM integration guide

This week’s roundup highlights five high-risk vulnerabilities affecting SD-WAN, load balancers, and AI platforms. Several flaws enable unauthenticated remote code execution or authentication bypass.

Key Findings

• Versa Concerto SD-WAN (CVE-2025-34026, CVE-2025-34027): Authentication bypass vulnerabilities allow attackers to gain unauthorized access to SD-WAN management interfaces, compromising network segmentation and control.
• Kemp LoadMaster (CVE-2024-7591): Remote Code Execution vulnerability enables attackers to execute arbitrary commands, potentially leading to full device compromise within enterprise load balancing environments.
• AnythingLLM (CVE-2024-0759): Server-Side Request Forgery (SSRF) flaw allows external attackers to force the LLM backend to make unauthorized internal network requests, potentially exposing sensitive internal resources.
• Anyscale Ray (CVE-2023-48022): Remote Code Execution vulnerability affecting distributed AI workloads, allowing attackers to execute arbitrary code on Ray cluster nodes.
• Server-Side Request Forgery (SSRF) - Generic & Obfuscated Payloads: Ongoing advancements in SSRF payload techniques observed, including obfuscation and expanded targeting of cloud metadata services and internal IP ranges.

Impact

These vulnerabilities expose critical infrastructure across networking, AI platforms, and SaaS integrations. Unauthenticated RCE and auth bypass flaws in Versa Concerto, Kemp LoadMaster, and Anyscale Ray allow full system compromise. AnythingLLM and SSRF payload variants expand attack surfaces into internal cloud resources, sensitive APIs, and metadata services, increasing risk of privilege escalation, data theft, and persistent access.

When you use the built-in build system that is part of Cloudflare Pages, the Build Image now includes Node.js v22. Previously, Node.js v18 was provided by default, and Node.js v18 is now end-of-life (EOL).

If you are creating a new Pages project, the new V3 build image that includes Node.js v22 will be used by default. If you have an existing Pages project, you can update to the latest build image by navigating to Settings > Build & deployments > Build system version in the Cloudflare dashboard for a specific Pages project.

Note that you can always specify a particular version of Node.js or other built-in dependencies by setting an environment variable.

For more, refer to the developer docs for Cloudflare Pages builds

You can now enable Polish with the webp format directly in Configuration Rules, allowing you to optimize image delivery for specific routes, user agents, or A/B tests — without applying changes zone-wide.

What’s new:

• WebP is now a supported value in the Polish setting for Configuration Rules.

This gives you more precise control over how images are compressed and delivered, whether you're targeting modern browsers, running experiments, or tailoring performance by geography or device type.

Learn more in the Polish and Configuration Rules documentation.

You can now debug, profile, view logs, and analyze memory usage for your Worker ↗ using Chrome Devtools ↗ when your Worker runs locally using the Cloudflare Vite plugin ↗.

Previously, this was only possible if your Worker ran locally using the Wrangler CLI ↗, and now you can do all the same things if your Worker uses Vite ↗.

When you run vite, you'll now see a debug URL in your console:

  VITE v6.3.5  ready in 461 ms

  ➜  Local:   http://localhost:5173/
  ➜  Network: use --host to expose
  ➜  Debug:   http://localhost:5173/__debug
  ➜  press h + enter to show help

Open the URL in Chrome, and an instance of Chrome Devtools will open and connect to your Worker running locally. You can then use Chrome Devtools to debug and introspect performance issues. For example, you can navigate to the Performance tab to understand where CPU time is spent in your Worker:

For more information on how to get the most out of Chrome Devtools, refer to the following docs:

• Debug code by setting breakpoints
• Profile CPU usage
• Observe memory usage and debug memory leaks

Users can now access significant enhancements to Cloudflare Gateway analytics, providing you with unprecedented visibility into your organization's DNS queries, HTTP requests, and Network sessions. These powerful new dashboards enable you to go beyond raw logs and gain actionable insights into how your users are interacting with the Internet and your protected resources.

You can now visualize and explore:

• Patterns Over Time: Understand trends in traffic volume and blocked requests, helping you identify anomalies and plan for future capacity.
• Top Users & Destinations: Quickly pinpoint the most active users, enabling better policy enforcement and resource allocation.
• Actions Taken: See a clear breakdown of security actions applied by Gateway policies, such as blocks and allows, offering a comprehensive view of your security posture.
• Geographic Regions: Gain insight into the global distribution of your traffic.

To access the new overview, log in to your Cloudflare Zero Trust dashboard ↗ and go to Analytics in the side navigation bar.

Users using Cloudflare's REST API to query their D1 database can see lower end-to-end request latency now that D1 authentication is performed at the closest Cloudflare network data center that received the request. Previously, authentication required D1 REST API requests to proxy to Cloudflare's core, centralized data centers, which added network round trips and latency.

Latency improvements range from 50-500 ms depending on request location and database location and only apply to the REST API. REST API requests and databases outside the United States see a bigger benefit since Cloudflare's primary core data centers reside in the United States.

D1 query endpoints like /query and /raw have the most noticeable improvements since they no longer access Cloudflare's core data centers. D1 control plane endpoints such as those to create and delete databases see smaller improvements, since they still require access to Cloudflare's core data centers for other control plane metadata.

We're excited to share that you can now use the Playwright MCP ↗ server with Browser Rendering.

Once you deploy the server, you can use any MCP client with it to interact with Browser Rendering. This allows you to run AI models that can automate browser tasks, such as taking screenshots, filling out forms, or scraping data.

Playwright MCP is available as an npm package at @cloudflare/playwright-mcp ↗. To install it, type:

npm i -D @cloudflare/playwright-mcp

yarn add -D @cloudflare/playwright-mcp

pnpm add -D @cloudflare/playwright-mcp

bun add -d @cloudflare/playwright-mcp

Deploying the server is then as easy as:

import { env } from "cloudflare:workers";
import { createMcpAgent } from "@cloudflare/playwright-mcp";

export const PlaywrightMCP = createMcpAgent(env.BROWSER);
export default PlaywrightMCP.mount("/sse");

Check out the full code at GitHub ↗.

Learn more about Playwright MCP in our documentation.

We have deployed an updated attack score model focused on enhancing the detection of multiple false positives (FPs).

As a result of this improvement, some changes in observed attack scores are expected.

With upgraded limits to all free and paid plans ↗, you can now scale more easily with Cloudflare for SaaS ↗ and Secrets Store ↗.

Cloudflare for SaaS ↗ allows you to extend the benefits of Cloudflare to your customers via their own custom or vanity domains. Now, the limit for custom hostnames ↗ on a Cloudflare for SaaS Pay-as-you-go plan has been raised from 5,000 custom hostnames to 50,000 custom hostnames.

With custom origin server -- previously an enterprise-only feature -- you can route traffic from one or more custom hostnames somewhere other than your default proxy fallback. Custom origin server ↗ is now available to Cloudflare for SaaS customers on Free, Pro, and Business plans.

You can enable custom origin server on a per-custom hostname basis via the API ↗ or the UI:

Currently in beta with a Workers integration ↗, Cloudflare Secrets Store ↗ allows you to store, manage, and deploy account level secrets from a secure, centralized platform your Cloudflare Workers ↗. Now, you can create and deploy 100 secrets per account. Try it out in the dashboard ↗, with Wrangler ↗, or via the API ↗ today.

All Cloudflare One Gateway users can now use Protocol detection logging and filtering, including those on Pay-as-you-go and Free plans.

With Protocol Detection, admins can identify and enforce policies on traffic proxied through Gateway based on the underlying network protocol (for example, HTTP, TLS, or SSH), enabling more granular traffic control and security visibility no matter your plan tier.

This feature is available to enable in your account network settings for all accounts. For more information on using Protocol Detection, refer to the Protocol detection documentation.

This week’s roundup covers nine vulnerabilities, including six critical RCEs and one dangerous file upload. Affected platforms span cloud services, CI/CD pipelines, CMSs, and enterprise backup systems. Several are now addressed by updated WAF managed rulesets.

Key Findings

• Ingress-Nginx (CVE-2025-1098): Unauthenticated RCE via unsafe annotation handling. Impacts Kubernetes clusters.
• GitHub Actions (CVE-2025-30066): RCE through malicious workflow inputs. Targets CI/CD pipelines.
• Craft CMS (CVE-2025-32432): Template injection enables unauthenticated RCE. High risk to content-heavy sites.
• F5 BIG-IP (CVE-2025-31644): RCE via TMUI exploit, allowing full system compromise.
• AJ-Report (CVE-2024-15077): RCE through untrusted template execution. Affects reporting dashboards.
• NAKIVO Backup (CVE-2024-48248): RCE via insecure script injection. High-value target for ransomware.
• SAP NetWeaver (CVE-2025-31324): Dangerous file upload flaw enables remote shell deployment.
• Ivanti EPMM (CVE-2025-4428, 4427): Auth bypass allows full access to mobile device management.
• Vercel (CVE-2025-32421): Information leak via misconfigured APIs. Useful for attacker recon.

Impact

These vulnerabilities expose critical components across Kubernetes, CI/CD pipelines, and enterprise systems to severe threats including unauthenticated remote code execution, authentication bypass, and information leaks. High-impact flaws in Ingress-Nginx, Craft CMS, F5 BIG-IP, and NAKIVO Backup enable full system compromise, while SAP NetWeaver and AJ-Report allow remote shell deployment and template-based attacks. Ivanti EPMM’s auth bypass further risks unauthorized control over mobile device fleets.

GitHub Actions and Vercel introduce supply chain and reconnaissance risks, allowing malicious workflow inputs and data exposure that aid in targeted exploitation. Organizations should prioritize immediate patching, enhance monitoring, and deploy updated WAF and IDS signatures to defend against likely active exploitation.

We’ve launched two powerful new tools to make the GraphQL Analytics API more accessible:

The new GraphQL API Explorer ↗ helps you build, test, and run queries directly in your browser. Features include:

• In-browser schema documentation to browse available datasets and fields
• Interactive query editor with autocomplete and inline documentation
• A "Run in GraphQL API Explorer" button to execute example queries from our docs
• Seamless OAuth authentication — no manual setup required

MCP Servers let you use natural language tools like Claude to generate structured queries against your data. See our blog post ↗ for details on how they work and which servers are available. The new GraphQL MCP server ↗ helps you discover and generate useful queries for the GraphQL Analytics API. With this server, you can:

• Explore what data is available to query
• Generate and refine queries using natural language, with one-click links to run them in the API Explorer
• Build dashboards and visualizations from structured query outputs

Example prompts include:

• “Show me HTTP traffic for the last 7 days for example.com”
• “What GraphQL node returns firewall events?”
• “Can you generate a link to the Cloudflare GraphQL API Explorer with a pre-populated query and variables?”

We’re continuing to expand these tools, and your feedback helps shape what’s next. Explore the documentation to learn more and get started.

In Cloudflare Workers, you can now attach an event listener to Request objects, using the signal property ↗. This allows you to perform tasks when the request to your Worker is canceled by the client. To use this feature, you must set the enable_request_signal compatibility flag.

You can use a listener to perform cleanup tasks or write to logs before your Worker's invocation ends. For example, if you run the Worker below, and then abort the request from the client, a log will be written:

export default {
	async fetch(request, env, ctx) {
		// This sets up an event listener that will be called if the client disconnects from your
		// worker.
		request.signal.addEventListener("abort", () => {
			console.log("The request was aborted!");
		});

		const { readable, writable } = new IdentityTransformStream();
		sendPing(writable);
		return new Response(readable, {
			headers: { "Content-Type": "text/plain" },
		});
	},
};

async function sendPing(writable) {
	const writer = writable.getWriter();
	const enc = new TextEncoder();

	for (;;) {
		// Send 'ping' every second to keep the connection alive
		await writer.write(enc.encode("ping\r\n"));
		await scheduler.wait(1000);
	}
}

export default {
  async fetch(request, env, ctx): Promise<Response> {
    // This sets up an event listener that will be called if the client disconnects from your
    // worker.
    request.signal.addEventListener('abort', () => {
      console.log('The request was aborted!');
    });

    const { readable, writable } = new IdentityTransformStream();
    sendPing(writable);
    return new Response(readable, { headers: { 'Content-Type': 'text/plain' } });
  },
} satisfies ExportedHandler<Env>;

async function sendPing(writable: WritableStream): Promise<void> {
	const writer = writable.getWriter();
	const enc = new TextEncoder();

	for (;;) {
		// Send 'ping' every second to keep the connection alive
		await writer.write(enc.encode('ping\r\n'));
		await scheduler.wait(1000);
	}
}

For more information see the Request documentation.

Earlier this year, we announced the launch of the new Terraform v5 Provider. Unlike the earlier Terraform providers, v5 is automatically generated based on the OpenAPI Schemas for our REST APIs. Since launch, we have seen an unexpectedly high number of issues ↗ reported by customers. These issues currently impact about 15% of resources. We have been working diligently to address these issues across the company, and have released the v5.5.0 release which includes a number of bug fixes. Please keep an eye on this changelog for more information about upcoming releases.

• Broad fixes across resources with recurring diffs, including, but not limited to:
  • cloudflare_zero_trust_gateway_policy
  • cloudflare_zero_trust_access_application
  • cloudflare_zero_trust_tunnel_cloudflared_route
  • cloudflare_zone_setting
  • cloudflare_ruleset
  • cloudflare_page_rule
• Zone settings can be re-applied without client errors
• Page rules conversion errors are fixed
• Failure to apply changes to cloudflare_zero_trust_tunnel_cloudflared_route
• Other bug fixes

For a more detailed look at all of the changes, see the changelog ↗ in GitHub.

• #5304: Importing cloudflare_zero_trust_gateway_policy invalid attribute filter value ↗
• #5303: cloudflare_page_rule import does not set values for all of the fields in terraform state ↗
• #5178: cloudflare_page_rule Page rule creation with redirect fails ↗
• #5336: cloudflare_turnstile_wwidget not able to update ↗
• #5418: cloudflare_cloud_connector_rules: Provider returned invalid result object after apply ↗
• #5423: cloudflare_zone_setting: "Invalid value for zone setting always_use_https" ↗

If you have an unaddressed issue with the provider, we encourage you to check the open issues ↗ and open a new one if one does not already exist for what you are experiencing.

If you are evaluating a move from v4 to v5, please make use of the migration guide ↗. We have provided automated migration scripts using Grit which simplify the transition, although these do not support implementations which use Terraform modules, so customers making use of modules need to migrate manually. Please make use of terraform plan to test your changes before applying, and let us know if you encounter any additional issues by reporting to our GitHub repository ↗.

• Terraform provider ↗
• Documentation on using Terraform with Cloudflare

This week's analysis covers four vulnerabilities, with three rated critical due to their Remote Code Execution (RCE) potential. One targets a high-traffic frontend platform, while another targets a popular content management system. These detections are now part of the Cloudflare Managed Ruleset in Block mode.

Key Findings

• Commvault Command Center (CVE-2025-34028) exposes an unauthenticated RCE via insecure command injection paths in the web UI. This is critical due to its use in enterprise backup environments.
• BentoML (CVE-2025-27520) reveals an exploitable vector where serialized payloads in model deployment APIs can lead to arbitrary command execution. This targets modern AI/ML infrastructure.
• Craft CMS (CVE-2024-56145) allows RCE through template injection in unauthenticated endpoints. It poses a significant risk for content-heavy websites with plugin extensions.
• Apache HTTP Server (CVE-2024-38475) discloses sensitive server config data due to misconfigured mod_proxy behavior. While not RCE, this is useful for pre-attack recon.

Impact

These newly detected vulnerabilities introduce critical risk across modern web stacks, AI infrastructure, and content platforms: unauthenticated RCEs in Commvault, BentoML, and Craft CMS enable full system compromise with minimal attacker effort.

Apache HTTPD information leak can support targeted reconnaissance, increasing the success rate of follow-up exploits. Organizations using these platforms should prioritize patching and monitor for indicators of exploitation using updated WAF detection rules.

42 new applications have been added for Zero Trust support within the Application Library and Gateway policy enforcement, giving you the ability to investigate or apply inline policies to these applications.

33 of the 42 applications are Artificial Intelligence applications. The others are Human Resources (2 applications), Development (2 applications), Productivity (2 applications), Sales & Marketing, Public Cloud, and Security.

To view all available applications, log in to your Cloudflare Zero Trust dashboard ↗, navigate to the App Library under My Team.

For more information on creating Gateway policies, see our Gateway policy documentation.

A new Access Analytics dashboard is now available to all Cloudflare One customers. Customers can apply and combine multiple filters to dive into specific slices of their Access metrics. These filters include:

• Logins granted and denied
• Access events by type (SSO, Login, Logout)
• Application name (Salesforce, Jira, Slack, etc.)
• Identity provider (Okta, Google, Microsoft, onetimepin, etc.)
• Users (chris@cloudflare.com, sally@cloudflare.com, rachel@cloudflare.com, etc.)
• Countries (US, CA, UK, FR, BR, CN, etc.)
• Source IP address
• App type (self-hosted, Infrastructure, RDP, etc.)

To access the new overview, log in to your Cloudflare Zero Trust dashboard ↗ and find Analytics in the side navigation bar.

You can now create Durable Objects using Python Workers. A Durable Object is a special kind of Cloudflare Worker which uniquely combines compute with storage, enabling stateful long-running applications which run close to your users. For more info see here.

You can define a Durable Object in Python in a similar way to JavaScript:

from workers import DurableObject, Response, WorkerEntrypoint

from urllib.parse import urlparse

class MyDurableObject(DurableObject):
    def __init__(self, ctx, env):
        self.ctx = ctx
        self.env = env

    def fetch(self, request):
        result = self.ctx.storage.sql.exec("SELECT 'Hello, World!' as greeting").one()
        return Response(result.greeting)

class Default(WorkerEntrypoint):
    async def fetch(self, request):
        url = urlparse(request.url)
        id = env.MY_DURABLE_OBJECT.idFromName(url.path)
        stub = env.MY_DURABLE_OBJECT.get(id)
        greeting = await stub.fetch(request.url)
        return greeting

Define the Durable Object in your Wrangler configuration file:

{
	"durable_objects": {
		"bindings": [
			{
				"name": "MY_DURABLE_OBJECT",
				"class_name": "MyDurableObject"
			}
		]
	}
}

[[durable_objects.bindings]]
name = "MY_DURABLE_OBJECT"
class_name = "MyDurableObject"

Then define the storage backend for your Durable Object:

{
	"migrations": [
		{
			"tag": "v1", // Should be unique for each entry
			"new_sqlite_classes": [ // Array of new classes
				"MyDurableObject"
			]
		}
	]
}

[[migrations]]
tag = "v1"
new_sqlite_classes = [ "MyDurableObject" ]

Then test your new Durable Object locally by running wrangler dev:

npx wrangler dev

Consult the Durable Objects documentation for more details.

You can now safely open email attachments to view and investigate them.

What this means is that messages now have a Attachments section. Here, you can view processed attachments and their classifications (for example, Malicious, Suspicious, Encrypted). Next to each attachment, a Browser Isolation icon allows your team to safely open the file in a clientless, isolated browser with no risk to the analyst or your environment.

To use this feature, you must:

• Turn on Allow users to open a remote browser without the device client in your Zero Trust settings.
• Have Browser Isolation (BISO) seats assigned.

For more details, refer to our setup guide.

Some attachment types may not render in Browser Isolation. If there is a file type that you would like to be opened with Browser Isolation, reach out to your Cloudflare contact.

This feature is available across these Email security packages:

• Advantage
• Enterprise
• Enterprise + PhishGuard

New categories added

Changes to existing categories

Refer to Gateway domain categories to learn more.

Hyperdrive has been approved for FedRAMP Authorization and is now available in the FedRAMP Marketplace ↗.

FedRAMP is a U.S. government program that provides standardized assessment and authorization for cloud products and services. As a result of this product update, Hyperdrive has been approved as an authorized service to be used by U.S. federal agencies at the Moderate Impact level.

For detailed information regarding FedRAMP and its implications, please refer to the official FedRAMP documentation for Cloudflare ↗.