New customer offer−20 %20% off the first 3 months when paid monthlyView plans
SilentShield

SilentShield and your WAF: what goes where

SilentShield is not a web application firewall (WAF). It answers who is sending a request: a person, a bot or an AI agent. A WAF answers what the request contains. This page shows how the two work together and what to configure so that neither blocks the other.

Two different questions

Your WAF: what is in the request?

A WAF checks requests for known attack patterns such as SQL injection, cross-site scripting or path traversal and blocks them by rule or signature. It looks at every request, no matter who sends it.

SilentShield: who is sending it?

SilentShield measures how a visitor behaves in the browser and returns a verdict to your server for a form submission or login: human, suspicious or bot. With agent rules, your server or edge also decides which AI agents may access your site. SilentShield does not search form fields for attack code.

What SilentShield does not do

  • It does not detect SQL injection, cross-site scripting or other attacks in the content of a request.
  • It does not close vulnerabilities in your application, theme or plugins (no virtual patching).
  • It does not absorb DDoS attacks. That is the job of your hosting provider or CDN.

Does SilentShield protect against SQL injection?

Not as a WAF would. If an automated tool sends an attack through a protected form, the submission can fail the bot check — because of how the sender behaves, not because of the attack. A person who types the attack by hand, or a request sent straight to your application without the form, is not stopped by it. Keep validating input on your server and keep your WAF.

Where each layer sits

On its way from the visitor to your application, a request typically passes these layers:

  1. DDoS protection and CDN, if you use them.
  2. WAF: filters attack patterns before the request reaches your application.
  3. Agent enforcer (optional): applies your agent rules on your server or at the edge, without calling SilentShield for each request.
  4. Your application with the SilentShield widget: the form sends the token in the hidden field behavior_nonce, and your server checks it with the SilentShield API before it processes the submission.

Configure your WAF for SilentShield

Widget traffic does not pass your WAF

The widget in the visitor's browser talks to the SilentShield hosts directly, not to your server. Your WAF only sees the submitted form. If it has strict rules for request bodies, allow the hidden field behavior_nonce.

Allow outgoing connections for the server-side check

Your server calls the SilentShield API to check the token. If outgoing connections are restricted, allow HTTPS (port 443) to these hosts:

https://api-us.silentshield.io
https://api-eu.silentshield.io
https://api.silentshield.io

Let requests from our servers through

Webhooks and the form test in the form rules reach your site from SilentShield's servers. Bot protection in front of your site, such as a bot-fight mode or a JavaScript challenge, often answers them with 403. The webhook is then not delivered, and the form test cannot read your page. Allow these requests for the affected paths. The form test identifies itself as SilentShieldScanner, our public form check as SilentShieldFormCheck. Both respect robots.txt: you can address each with its own group, such as User-agent: SilentShieldFormCheck; a group for SilentShieldScanner applies to the public form check as well.

Avoid a second challenge

If your WAF or CDN shows its own bot challenge on the page with the form, visitors may have to pass two checks. Exclude the protected pages from that challenge, or keep it for paths without SilentShield.

Related pages: