Cloudflare Security Review · Read-only · Checked by a person

Is your Cloudflare account set up the way you think it is?

We review your websites, email records, API tokens, Worker secrets, and AI settings. A person checks every finding and explains it in plain English.

See a sample report The sample uses made-up, redacted example data.

Read-only access.A 7-day token you can revoke.No changes without your written OK.

The problem, in plain English

Cloudflare accounts grow over time. Someone adds a zone, a Worker, an AI Gateway, or an API token. Often nobody writes it down.

Cloudflare's built-in Security Insights helps. It does not show everything.

Website and DNS settings are easy to see. Account settings are easy to miss.

Most findings are cleanup items. That is normal. The goal is a clean account that you understand.

Here are a few examples:

  • An API token that can make changes, never expires, and works from any IP address.
  • A Worker variable that stores a key in plain text.
  • An AI Gateway with no login and no spend limit.
  • An AI agent or MCP server that sits outside Cloudflare Access.

You won't see any of these by visiting your website. Our report shows where each one is and how to fix it.

What the review checks

We run 41 configuration checks across your Cloudflare account. Then we run 9 outside-in verifications from the public internet.

Websites, DNS, and perimeter

9
  • TLS versions below 1.2, and SSL set to Off or Flexible.
  • Always Use HTTPS and HSTS settings.
  • DNSSEC off or stuck in pending.
  • A missing CAA record, which controls who may issue your certificates.
  • Wildcard DNS records.
  • Unproxied records that show your origin server's IP address.
  • Whether a managed WAF ruleset is on.

Email spoofing protection

13
  • SPF, DMARC, and DKIM for Microsoft 365, Google, Zoho, and Proton.
  • MTA-STS and TLS-RPT for mail in transit.
  • Parked domains that send no mail and can still be spoofed.

Account access and API tokens

5
  • Account members without two-factor login.
  • More than 2 Super Administrators.
  • Tokens that can make changes and never expire.
  • Tokens that can make changes and have no IP limit.
  • Tokens unused for 90 days or more.

AI Gateways, Workers, MCP, and chat

11
  • AI Gateway authentication, logs, guardrails, and spend limits.
  • MCP or agent hosts that are not fully behind Cloudflare Access.
  • AI Workers that are also open on workers.dev.
  • Public chat pages with no Turnstile.

Secrets and AI spend

3
  • Plain-text Worker variables whose values look like keys.
  • Plain-text Worker variables named like secrets.
  • Workers AI with no spend ceiling.
  • We record the variable name. We never record the value.

Outside-in verification

9
  • Public SPF, DMARC, MX, and CAA records match what Cloudflare holds.
  • DNSSEC validates end to end.
  • The TLS versions each host really accepts.
  • HTTP redirects to HTTPS, and the HSTS header is sent.
  • Optional: DKIM signing proven from a real message header.

Also in every report

  • A list of your AI hosts, Workers, and AI Gateways for you to confirm.
  • A list of anything our read-only token could not see. We mark it "could not see." We never mark it "clean."
  • DNS records that are unproxied on purpose, such as Microsoft 365 records. We list them apart so they don't count as problems.

What free tools miss

Free tools are a good place to start. We tested two popular free scanners and Cloudflare Security Insights on our own accounts.

The account layer. Our review found issues with API tokens, Worker secrets, and AI spend limits. Neither free scanner found these on our test account.

A person checks every finding. Scanners give you a list. We check each finding by hand before it reaches you. Then we explain what it means and how to fix it.

The numbers, with the fine print.

  • Our review found all 17. One free scanner found 8, plus 1 in part. The other found 5.
  • On one account we tested earlier, Security Insights showed 6 suggestions. Our review found 30 items. 22 of those were not in Security Insights.
  • Each result comes from a single account. Your account will differ.

On our test account, we tracked 17 real issue types.

Our review
17
One free scanner
8 +1 in part
The other
5
We built that list of 17 from our own findings, so it favors us.

What you get

A read-only review.

We look. We do not touch.

A plain-English report, checked by a person.

Each item shows what we found, the evidence, and how to fix it. It also shows how risky the fix is and how to undo it.

Context you can trust.

Each item notes whether Security Insights also shows it. It also notes whether our outside checks confirm it.

A 30-minute walkthrough call.

We go through the report with you and answer questions. You decide what to fix and when.

See a sample report The sample uses made-up, redacted example data.

Optional fixes add-on

Want help with the fixes? We can guide you through a pre-agreed set of simple fixes, up to 3 analyst hours.

Prefer to make the fixes yourself? The report tells you how.

  • We write a plan first.
  • You approve each exact change in writing before your team makes it.
  • We walk your team through each fix on a live screen-share, with step-by-step instructions. We never need write access.
  • Every change comes with an undo step.
  • We check each change afterward to confirm it worked.
  • Complex WAF work, migrations, and long DMARC rollouts are quoted separately.

How it works

  1. Sign a short letter. The scope and authorization letter says exactly what is in scope.
  2. Create a read-only token. We send step-by-step instructions. The token expires in 7 days and can be limited to our IP address. You can revoke it at any time.
  3. We collect your settings once. We use read-only calls to Cloudflare and make no changes.
  4. We run light outside checks. These are public DNS lookups, a few TLS handshakes, and one web request per proxied host. That is about the same as a browser visiting your homepage. We run no scans, no load tests, and no login attempts.
  5. You get the report and the call. We walk you through every finding in 30 minutes.
  6. Optional: fix what matters. We guide your team through changes you approve, or you make them on your own.

What this review is not

This is a configuration review. It is not a penetration test. We do not attack your apps or try to log in.

What it doesn't cover:

  • Your application code and MCP tool definitions.
  • AI tools your staff use outside Cloudflare.
  • Other clouds, Microsoft 365 app secrets, and GitHub.
  • R2, D1, and Zero Trust Gateway policies, which we don't check yet.

Honest limits:

  • Our DKIM check finds selectors in DNS. It proves mail is signed only if you send us a real message header.
  • Our AI checks look at configuration. They do not prove that app login or Turnstile works in your code.
  • A clean result does not mean your systems cannot be compromised.

Pricing

Cloudflare Security Review

$1,200
one time
  • Covers 1 Cloudflare account and 1 mail tenant.
  • Covers up to 5 zones, 10 public hosts, 10 Workers, and 2 AI Gateways.
  • 41 configuration checks and 9 outside-in verifications.
  • Plain-English report, checked by a person.
  • 30-minute walkthrough call.
  • Larger accounts are quoted separately.

Fixes add-on

$750
  • Guided help with a pre-agreed set of simple fixes, up to 3 analyst hours.
  • Your written approval of each change, undo steps, and a check after.

All prices in US dollars.

For MSPs who resell

Resell the review at a wholesale price of $850 per review. Your first review is $600, then $850 wholesale per review. You handle the authorization letter, the client's token, and the client call. We run the review and deliver the report.

Scope per review: 1 Cloudflare account, 5 zones, 10 public hosts, 10 Workers, 2 AI Gateways, and 1 mail tenant. Larger accounts are quoted separately.

Access model:

  • A read-only Cloudflare API token with a 7-day expiry.
  • Client IP filtering where possible. The client can revoke the token at any time.
  • We never request a token with edit permissions.

Collection:

  • One pass of read-only Cloudflare API calls.
  • Outside-in: public DNS for SPF, DMARC, MX, CAA, and the DNSSEC chain.
  • TLS handshakes per host to confirm the versions each one accepts.
  • One HTTP request per proxied host to check the HTTPS redirect and HSTS header.
  • No scanning, load testing, or authentication attempts.

Account-layer coverage:

  • API tokens with edit permissions and no expiry or no IP condition.
  • Tokens unused for 90 days or more, or never used.
  • Members without 2FA, and more than 2 Super Administrators.
  • Plain-text Worker variables that match key patterns or have secret-like names. We record names only, never values.
  • AI Gateway authentication, logging, guardrails, DLP, and gateway-wide rate or spend limits.
  • Workers AI with no spend ceiling.
  • MCP and agent hosts not fully behind Cloudflare Access.
  • AI Workers also reachable on workers.dev, which bypasses Access and WAF.
  • AI Security for Apps status, and whether any WAF rule uses it.
  • Authenticated Origin Pulls on AI host origins.

Report output:

  • Each finding has evidence, fix steps, fix risk, and rollback steps.
  • Each finding notes Security Insights overlap and outside-in confirmation.
  • Gaps in token visibility are marked "could not see."
  • Records unproxied by design, such as Microsoft 365, are listed separately.

Known gaps today:

We don't check SSL "Full" versus "Full (strict)" yet. R2, D1, and Zero Trust Gateway policies are not covered yet.

Next step: Talk to Ed about a partner setup.

FAQ

Cloudflare Security Insights is free. Why pay for this?

Security Insights is useful, and it comes with every Cloudflare plan, even Free. It does not include an analyst-led report or help with fixes. On one account we tested, it showed 6 suggestions. Our review found 30 items, and 22 of those were not in Security Insights. That was one account, so yours will differ. Our report also notes whether Security Insights shows each item.

We already run a free scanner. Do we need this?

You can keep running it. On our test account, one popular free scanner matched us on basic zone settings. It also flags SSL "Full" versus "Full (strict)," which we don't check yet. Neither free scanner found our API token, Worker secret, or AI spend limit issues on that account. Our review covers that account layer. A person also checks each finding before it reaches you.

Why would we give you an API token?

The token is read-only, so it cannot change anything. You create it yourself, with a 7-day expiry and an IP limit if possible. You can revoke it at any time. We never ask for a token that can make changes. We also sign a scope and authorization letter before we start.

Is this a penetration test?

No. It is a configuration review. We do not attack your apps, run scans, or try to log in.

What do the outside checks do to our website?

They are light. We run public DNS lookups, a few TLS handshakes, and one web request per proxied host. That is about the same as a browser visiting your homepage.

Will you help fix the problems you find?

Only if you want us to. With the fixes add-on, you approve each change in writing. We walk your team through each fix on a live screen-share, with step-by-step instructions. We never need write access. Each change has an undo step and a check afterward. You can also make the fixes yourself using the report.

We use Microsoft 365. Does that matter?

Our email checks cover Microsoft 365, Google, Zoho, and Proton. Some Microsoft 365 records are unproxied on purpose. We list those apart so they don't count as problems.

Is Secure Data Solutions part of Cloudflare?

No. "Works with Cloudflare" describes the platform we review. Secure Data Solutions is not affiliated with or endorsed by Cloudflare, Inc.

Eddie Espino, founder of Secure Data Solutions

About Ed

Eddie Espino is the founder of Secure Data Solutions, based in West Palm Beach, Florida. He has more than 20 years of experience in firewall, network security, and penetration testing work.

Secure Data Solutions does hands-on security work for small and mid-size businesses. We also work with the MSPs that serve them. We work read-only by default and explain what we find in plain English. Nothing changes without your written OK.

See what is really in your Cloudflare account.

Email Ed to get started. We will send you the authorization letter and the token steps.