Skip to content

Cloudflare

Integrate your Risk Register and Incident Register with Cloudflare. A risk integration imports Security Center insights, Zero Trust CASB findings, and Page Shield malicious scripts as risks. An incident integration imports WAF attack campaigns and Brand Protection matches as incidents. Both use the same kind of connection, so one connection can serve both. You can also use separate connections, for example with API tokens scoped to different permissions.

Cloudflare Source Register Adversarial Source Class
Security Center insights Risks Attack Surface Monitoring Configuration
CASB findings Risks Third-party Review Third-party
Page Shield malicious scripts Risks Attack Surface Monitoring Vulnerability
WAF Incidents Web Application Firewall (WAF) —
Brand Protection Incidents Managed Detection and Response (MDR) —
  • Opened By: The integration’s integration user, which has the same name as the integration.

Create a Cloudflare API token (a user token or an account token) with Read access for each source you plan to import:

Source Token Permissions
Security Center insights Account › Account Security Center Insights › Read
CASB findings Account › Zero Trust › Read
Page Shield Zone › Zone › Read and Zone › Client-side security › Read
WAF Zone › Zone › Read and Zone › Analytics › Read
Brand Protection Account › Cloudforce One › Read

You also need your Cloudflare Account ID, the 32-character ID shown on the account’s overview page in the Cloudflare dashboard.

  1. In Adversarial, go to Settings > Integrations and select Add.
  2. Select RSK in the Cloudflare row to import risks, or INC to import incidents.
  3. Enter an Instance Name, such as Cloudflare Risks.
  4. Open Connection and select Add a connection…. Enter a Connection Name and the Account ID, paste the API token into Api key, then select Save Connection.
  5. Under Sources, tick the sources to import. None are ticked by default, and the integration can’t be enabled until at least one is.
    • Risks: Security insights, CASB, and Page Shield.
    • Incidents: WAF and Brand Protection. When WAF is ticked, two more settings appear:
      • Minimum requests (default 10) — how many requests one source must send, blocked or not, before it is imported as an incident. It also needs at least one security event.
      • Lookback hours (default 24, at most 720) — how many hours of activity each incident describes. Cloudflare limits the analytics window by plan: 24 hours on Free and Pro, 72 on Business, and 720 on Enterprise. Set it no higher than your plan allows.
  6. Select Save.

Cloudflare integration configuration

To import both risks and incidents, add a second integration from the other column. Choose the same connection, or add a separate connection with its own API token if you want to scope each token’s permissions to the products it imports.

Test on the connection checks that the API token is valid and active, and that the Account ID is 32 hexadecimal characters. It does not check that the token can reach that account, so a user token can pass with the wrong Account ID.

Test on an integration previews the first records each ticked source returns. If the token is missing a source’s permission, the preview names the permission it needs.

Each sync reads the full list from Cloudflare. Adversarial creates one risk per:

  • Security Center insight, including archived insights, which import as Closed.
  • CASB finding per connected SaaS integration in Cloudflare One. Only posture findings are imported; DLP content findings are not.
  • Page Shield script per zone, for scripts Cloudflare flags as malicious only.
Cloudflare State Adversarial Status
Security insight open New
Security insight archived or resolved Closed
CASB finding open, with active instances New
CASB finding ignored, or with no active instances left Closed
Page Shield malicious script New

Cloudflare severity maps to Adversarial Initially Reported Urgency (IRU):

Source Cloudflare Severity Adversarial IRU
Security insights Critical Critical
Security insights Moderate Medium
Security insights Low Low
CASB Critical, High, Medium, Low Same level
Page Shield URL or domain reported malicious by threat intelligence Critical
Page Shield Flagged as malicious by Cloudflare’s script classifier High

For CASB, a severity an administrator has overridden in Cloudflare is used in place of the original. Page Shield scripts are imported only when they are flagged by threat intelligence or score 10 or lower (on Cloudflare’s 1–99 scale) for JavaScript integrity, malware, cryptomining, or Magecart. Cloudflare reports these signals only on zones whose plan includes them, so other zones import no scripts.

Adversarial Field Security Insights CASB Page Shield
Title Insight class and subject, such as “Always use https not enabled (example.com)” “{finding} — {integration}”, such as “Slack File Publicly Accessible — Corp Slack” “Malicious script on {zone}: {script host}”
Description Link to the insight in Security Center; class, type, severity, subject, zone, and detection method; Cloudflare’s remediation text Link to the finding’s active instances in Cloudflare One; finding description; vendor, integration, category, severity, active and archived instance counts; a table of up to 100 active instances with each asset, its type, and when it was detected, and a count of any not listed; threat, impact, and compliance frameworks Link to Cloudflare; script URL, host, zone, first and last seen; threat signals and scores; up to 10 pages that load the script
Remediation Task Cloudflare’s remediation text Cloudflare’s remediation guide Fixed text: confirm the script is expected on the listed pages, and remove it or block it with a content security rule if not
Discovered Date When Cloudflare first detected the insight The earliest detection date among the finding’s instances, including hidden ones; the import date if it has none When Cloudflare first saw the script
Closed Date The insight’s last scan, once closed The finding’s latest occurrence, once closed Not set
IRU Via severity mapping above Via severity mapping above Via severity mapping above
Source Always “Attack Surface Monitoring” Always “Third-party Review” Always “Attack Surface Monitoring”
Class Always “Configuration” Always “Third-party” Always “Vulnerability”

Title, Description, Remediation Task, IRU, Source, and Class are updated from Cloudflare on every sync, so edits to them in Adversarial are overwritten. The Discovered Date is also refreshed from Cloudflare on each sync. Cloudflare provides no assignee, likelihood, or impact.

When the AI switch is on, new risks are scored on import. A later sync does not rescore them, even when a CASB finding’s instance counts change its description.

Severity for both incident sources is set by Adversarial from what Cloudflare reports, not by AI scoring, so the AI switch has no effect on Cloudflare incidents.

Brand Protection imports matches from the string and logo queries saved in your Cloudflare dashboard. With no saved queries, nothing is imported, and the integration’s Test says so. Logo queries require a Cloudflare Enterprise plan.

  • One incident per lookalike domain, however many string queries matched it, titled “Lookalike domain: {domain}”.
  • One incident per logo match, titled “Logo impersonation: {domain}”.
Match Adversarial Severity Adversarial Status
Logo match (a page serving your logo) SEV-3 New
Lookalike domain SEV-4 New; Closed once the domain is dismissed on every query that matched it

When a match is no longer listed, for example because its saved query was deleted, the nightly full sync closes its incident and adds a comment. A dismissal doesn’t close an incident whose status a person set last. The severity reasoning explains the match kind: a logo match means the impersonation is live, while a lookalike domain may not be in use yet.

Cloudflare Field Adversarial Field Notes
Domain Title See title formats above
(multiple fields) Description Link to Brand Protection; domain; whether a string or logo query found it, and the query’s tags; registrar; first seen; scan status; logo similarity; URL Scanner report link
First seen (match time for logos) Detected Date Import time when Cloudflare has no date
First seen Occurred Date
(static) Source Always “Managed Detection and Response (MDR)”

On later syncs, the title, description, and dates (when Cloudflare has them) are refreshed. Severity is set on import only, so a severity you change in Adversarial stays. Status changes only when Cloudflare dismisses a domain, which closes it, unless a person set the status last.

The WAF source groups each attacker’s activity across every active zone into one incident per source address: one per IPv4 address, or one per IPv6 /64. A source is imported once, within Lookback hours, it has at least one security event and at least Minimum requests in total, counting allowed traffic. Allowed, skipped, bypassed, and rewritten requests, and solved challenges, are not security events. Counts are Cloudflare’s sampled estimates.

Each incident’s title states what the WAF analytics show, with the address and its top country:

Title Adversarial Severity Status on Import When
WAF blocked 203.0.113.7 (US) SEV-4 Closed No request got a response that needs review
WAF blocked 203.0.113.7 (US), some requests succeeded SEV-4 Review Some requests got a response that needs review, as listed below
False positive: WAF blocked 203.0.113.7 (US) SEV-5 Review Requests got 2xx on paths that returned 401 or 403 to other clients, or 2xx writes, and no request matched a WAF attack rule. This can be a real user

A response needs review when the source got any of these:

  • 2xx on paths that returned 401 or 403 to other clients.
  • 2xx on POST, PUT, PATCH, or DELETE requests.
  • 2xx on paths named like sensitive files, such as version-control metadata, environment files, or backups.
  • 2xx on requests that a WAF attack rule in log mode matched.
  • An average 2xx response over 100 KB on a restricted path, from a source that also matched WAF attack rules.
  • 5xx responses from the origin to at least 50 of its requests, and to at least 10% of them.

Severity comes from Cloudflare’s request and response analytics, never from scoring. It is never higher than SEV-4, because the analytics don’t show intent or impact. Raise it yourself once you confirm impact. The severity reasoning lists the facts that set the severity, and for SEV-4 it ends “Request and response analytics do not show impact.”

Each new incident gets a comment that starts “Closed:” or “In Review:” and says why, with a line on the source’s recent activity. Threat objectives that fit the matched WAF rules, such as Data Disclosure, Fraud, or Sabotage, are linked to a new incident at Moderate relevance. A later sync doesn’t link threat objectives, so a link you remove stays removed.

Adversarial Field Notes
Title As in the table above
Description The source’s requests across the account’s zones within Lookback hours; the actor (addresses, networks, countries, user agents, claimed bots, zones); the WAF rule matches; the traffic that got through, by destination; and, for a false positive, the rules and paths it matched with recent Ray IDs. When a source’s traffic reaches Cloudflare’s analytics row limit, the description says that some requests may be missing
Detected Date When the incident was first imported
Contained Date Same as Detected Date when the WAF blocked the source and the incident is Closed. Not set for an incident in Review
Severity As in the table above
Source Always “Web Application Firewall (WAF)”

Occurred and Responded dates are not set.

Incident State What a Sync Does
Open Refreshes the title and description. Severity only rises. An incident still in New moves to Closed or Review, with a comment that starts “Closed:” or “Moved to Review:”. When it moves to Closed, an empty Contained Date is set to the Detected Date.
Closed, source quiet since the last sync Nothing
Closed, source back and needing review Reopens to Review, with a comment that starts “Reopened to Review:”, if the severity rose or a person did not close it. The reopen clears the Contained Date, unless a person set it.
Closed, otherwise Adds a comment that starts “Active again:”, at most once a day. If the source needs review but a person closed the incident, the comment says “A person closed this incident, so it stays closed.”