skip to content

In Cloudflare's WAF, how do the Cloudflare Managed Ruleset and the Cloudflare OWASP Core Ruleset differ in how they decide to act?

level: middleimportance: should knowfreq 42%

answer

  1. signatures versus a running total
  2. per-rule action versus one final rule
  3. paranoia level picks which rules run
  4. Low 60, Medium 40, High 25

basics

~20 s

The Cloudflare Managed Ruleset matches signatures, and each matching rule applies its own action. The OWASP Core Ruleset adds the scores of matching rules into a threat score and acts only when that total reaches the threshold.

solid answer

~50 s

Both rulesets run in the `http_request_firewall_managed` phase, after custom rules and rate limiting rules, and each is deployed by a rule with the *Execute* action. The **Cloudflare Managed Ruleset** is Cloudflare's own signature set. It includes fast virtual patches for new CVEs, every rule has a default action set by its severity, some rules are disabled by default, and tags named after the software they protect let you turn on the rules for your stack. The **Cloudflare OWASP Core Ruleset** is Cloudflare's implementation of the OWASP ModSecurity Core Rule Set 3.3.0. Its individual rules only add points. The paranoia level (PL1 by default, up to PL4) decides which rules run. The score threshold (*Low* 60, *Medium* 40 by default, *High* 25) decides when the action on the last rule, `949110: Inbound Anomaly Score Exceeded`, fires. Both rulesets need a Pro plan or higher.

go deeper

for a junior

Recall that the Managed Ruleset is signature-based with an action per rule, while the OWASP ruleset sums scores and acts once through rule 949110.

for a middle

Explain paranoia level versus score threshold, the Low 60, Medium 40, High 25 values, and the ruleset, tag and rule levels where each ruleset is overridden.

for a senior

Show you read a 949110 event by opening Additional logs, and that you tune one Managed Ruleset rule without touching the others.

for a principal

Weigh whether the OWASP ruleset earns its false-positive cost on top of the Managed Ruleset and attack score for a given application.

## Where both rulesets sit Cloudflare's **managed rulesets** are rule collections that Cloudflare writes and updates. You deploy them on a zone instead of writing the rules yourself. Two of them do the work of a classic web application firewall: the **Cloudflare Managed Ruleset** and the **Cloudflare OWASP Core Ruleset**. They share a lot of plumbing: - Both run in the `http_request_firewall_managed` **phase**. That phase comes after HTTP DDoS Attack Protection, after WAF custom rules and after rate limiting rules, so a request that an earlier phase has already blocked or challenged never reaches them. - Each is switched on by a rule with the *Execute* action in the zone's entry point ruleset for that phase. Turning a ruleset on in the dashboard (Security **Settings**, filtered by **Web application exploits**) creates that rule for you. - At the zone level each managed ruleset can be deployed once. Its **scope** (a filter expression on the Execute rule) can narrow it to part of the traffic. - Both are available on Pro, Business and Enterprise. Free zones get the **Cloudflare Free Managed Ruleset**, a smaller subset that is deployed by default. Where they differ is the decision model, and that difference drives everything about how you tune them. ## The Cloudflare Managed Ruleset: each signature carries its own action The Cloudflare Managed Ruleset is written by Cloudflare's security team. It matches **known attack signatures** and gets **virtual patches** for newly disclosed vulnerabilities, often within hours. Each rule stands alone: - Every rule has a **default action** chosen by Cloudflare according to the rule's severity. - **Some rules are disabled by default** to keep false positives down. The docs advise against enabling every rule except for a proof of concept. - Rules carry **tags**, many named after the content management system or framework a rule protects, and Cloudflare recommends enabling the rules whose tags match your technology stack. When one rule matches, that rule's action applies. A false positive therefore traces back to a single rule ID, and you fix it for that rule. ## The OWASP Core Ruleset: a running total and one action The Cloudflare OWASP Core Ruleset is Cloudflare's implementation of the OWASP ModSecurity Core Rule Set, version 3.3.0. Its rules work as one unit: 1. Each matching rule adds its score to the request's **threat score** (also called the anomaly score). 2. The **paranoia level** decides which rules are evaluated at all. Setting PL3 enables the PL1, PL2 and PL3 rules, and PL4 rules are not evaluated. PL1 is the default. 3. The **score threshold** decides when the total is high enough: *Low – 60 and higher*, *Medium – 40 and higher* (the default) or *High – 25 and higher*. 4. If the total reaches the threshold, the action configured on the ruleset's **last rule**, `949110: Inbound Anomaly Score Exceeded`, runs. The available actions are Block (the default), Log, Non-Interactive Challenge, Managed Challenge and Interactive Challenge. Watch the naming: *High* means high sensitivity, not a high number. It needs the fewest points (25) and blocks the most requests. ## Side by side | | Cloudflare Managed Ruleset | Cloudflare OWASP Core Ruleset | | --- | --- | --- | | Detection model | Signatures, each judged alone | Rule scores summed into a threat score | | Where the action lives | On each rule (default by severity) | On rule `949110` only | | Main dials | Rule and tag status and action | Paranoia level, score threshold, OWASP action | | Rule-level dashboard override | Action and status | Status only | | False-positive profile | Low by design | Prone to false positives, needs tuning | ## What this means when tuning - In **Security Events**, a request mitigated by the OWASP ruleset shows rule `949110`, not the rules that added the points. The contributing rules and their scores are under **Additional logs** in the event's details. - Raising the paranoia level or picking a *High* threshold makes the whole OWASP ruleset stricter at once. A Managed Ruleset rule is tuned on its own. - Larger request bodies give OWASP rules more text to match. That pushes the total up and makes a false positive more likely. - Cloudflare's own docs warn that the OWASP Core Ruleset is prone to false positives and adds only marginal benefit on top of the Cloudflare Managed Ruleset and WAF attack score. A team that deploys it should plan to monitor and adjust it. The general trade-off of paranoia levels and anomaly thresholds, on any Core Rule Set deployment, belongs to the lesson on judging a WAF's rule sets. Here the point is how Cloudflare exposes those dials and where it records the result.

  • How do you set the OWASP ruleset's threshold and action through the Rulesets API rather than the dashboard?
    Add a rule override, inside the `overrides` of the Execute rule that deploys the ruleset, for its last rule (`949110: Inbound Anomaly Score Exceeded`). Set `score_threshold` to an integer such as 60 and `action` to a value such as `managed_challenge` or `log`. The paranoia level is set differently: with tag overrides that disable `paranoia-level-3` and `paranoia-level-4`, for example, to run at PL2.
  • Do you need the Cloudflare Free Managed Ruleset once the Cloudflare Managed Ruleset is deployed?
    No. The Free Managed Ruleset is a subset aimed at high-impact, widely exploited vulnerabilities and is deployed by default on Free zones. The Cloudflare Managed Ruleset covers more ground, so on a paid zone that runs it the free subset adds nothing.

  • Each OWASP Core Ruleset rule blocks on its own, like a signature.
  • The High score threshold needs the most points before it acts.
  • Setting PL4 runs only the PL4 rules.
  • Enabling every Cloudflare Managed Ruleset rule is the safest production setting.
  • In Security Events, an OWASP block names the rule whose pattern matched.

Assumes the Cloudflare OWASP Core Ruleset as of 2026-10-10, which implements Core Rule Set 3.3.0; Cloudflare tracks upstream releases, so the base version can move.