Semantic WAF

Protect applications with a reverse-proxy WAF that evaluates request structure, attack signals, and the policy for each project. A strong detection blocks the request and a weak one is logged.

RequestHTTPSQL injectionOne of 20 enginesXSSOne of 20 enginesSSRFOne of 20 enginesVerdictStrong or weakRequestHTTPSQL injectionOne of 20 enginesXSSOne of 20 enginesSSRFOne of 20 enginesVerdictStrong or weak
  • Request that passes
  • Request blocked

What it provides

  • 20 semantic engines, listed in the detection engine reference, covering SQL, XSS, commands, SSRF, JWT and GraphQL abuse, injection families, and other attack patterns.
  • Each engine set to Block, Log only or Off, per project and per path. Twelve engines also have a sensitivity level: Relaxed, Balanced or Strict.
  • Path policies and request-size limits.
  • Decision records with category, match detail and context for investigation.

How to use it

  1. Route application traffic through a connected WAF agent and configure its upstream.
  2. Set each engine to Block, Log only or Off, adjust the sensitivity level where needed, and set the relevant path policies for the application.
  3. Observe decisions in log mode, review legitimate traffic, and enable enforcement for the validated policy.
Route trafficThrough the WAF agentSet policyEngines, sensitivity, pathsLog, then enforceAfter reviewing trafficRoute trafficThrough the WAF agentSet policyEngines, sensitivity, pathsLog, then enforceAfter reviewing traffic
  • Order of the steps

Configuration and operation

Configure the project’s WAF settings. Connected agents receive configuration bundles over gRPC with mutual TLS. The semantic engines report what they find: a strong detection blocks the request and a weak one is logged.

Limits

Detection depends on the payload, available request data, and policy. No WAF guarantees that every attack is detected or every legitimate request is allowed. Tune with representative traffic and retain application-side validation.