AI semantic detection
Detects SQL injection, XSS, command injection and path traversal, including obfuscated variants.
RAIMON雷門Early access
An AI model judges whether each request is an attack. Every block comes with a stated reason.
Unfamiliar attack payloads are judged by the model. Published CVEs are blocked by rules.
Logs record the attack type and the reason for the verdict, for example "SQL injection obfuscated with comments".
Monitor mode only logs verdicts and does not affect traffic. Switch to block mode after you have reviewed them.
After the trial, billing is monthly by plan and you can stop anytime.
TRY IT
Illustrative. This uses simplified rules in your browser, not the actual model.
HOW IT WORKS
Five steps, from receiving the request to the verdict.
Undo layered encoding and obfuscation to see the original content.
Check whether the content is executable SQL, script or a shell command.
Give a risk score from 0 to 1 based on the request's structure and context.
Stop known vulnerabilities with rules, and check request rate and bot signals.
Reach a verdict and leave a reason you can read.
This flow is an architectural sketch. See the official documentation for details.
COMPARE
For published vulnerabilities, rules are the most precise. Unseen variants need a model. RAIMON uses both.
| Rules-only WAF | RAIMON AI-WAF | |
|---|---|---|
| Unseen attack variants | Not stopped until rules are updated | Judged by what the content means, without signature matching |
| Blocking legitimate users | A keyword match can cause a false block | Reads context, so normal text with quotes gets through |
| Day-to-day upkeep | Rules need constant tuning | Works with the defaults |
| Reason for the verdict | Shows only the ID of the rule that matched | States which kind of attack it was |
| Known vulnerabilities (CVEs) | Blocked by rules | Also blocked by rules, with virtual patching |
FEATURES
Feature list includes planned items
Detects SQL injection, XSS, command injection and path traversal, including obfuscated variants.
Covers the main web risks listed in the OWASP Top 10.
Checks the structure of JSON and API requests and blocks abnormal calls.
Tells search engines and real users apart from malicious crawlers and credential stuffing tools.
Limits request rate per source to slow brute force and HTTP flood attacks.
When a high-risk CVE is published, block it at the WAF first and schedule the fix after.
Set allow or block conditions by path, source and header.
Every block records the request, the score and the reason.
SETUP
Once enabled from the panel's plugins, the sites on that server connect to RAIMON in the cloud. It is the same AI-WAF as the standalone product.
Install the panelPoint your domain's DNS at RAIMON so traffic passes through AI-WAF first. The steps are below.
Start the 14-day trialThree steps for standalone use
# DNS — Samplewww.example.com. CNAME example-com.waf.raijin.jp. # Modemode: monitor # log only, no blockingmode: block # start blocking
PRICING
After the trial you're billed monthly by plan, in US dollars.Prices and plan contents are samples
FAQ
A traditional WAF matches requests against rules for known attack signatures. An AI WAF also uses a model to analyze what the request content means, so it recognizes variants that no rule covers and is less likely to block something because of a keyword match. RAIMON uses both.
Any WAF can misjudge. That is why we suggest starting in monitor mode, which logs verdicts without blocking. Turn blocking on once you've confirmed there are no problems.
No code changes. For standalone use, you only change a DNS record. If you already run the RAIJIN panel, turn it on from the panel's plugins.
Email us your current setup and we'll reply with the recommended way to connect.
After the 14-day trial, you're billed monthly in US dollars for the plan you chose. You can stop anytime.
The nodes are distributed worldwide. Specific locations will be announced before the official launch.
Enable it from the panel, or connect it on its own by changing DNS. The trial runs for 14 days.