Threat protection
Besides the WAF, NullWard can block bad traffic by IP address, reputation and behaviour, challenge bots, and scan uploads for malware. These features are under Threat Intelligence and Malware Detection in the sidebar, and are turned on per service.
Blocklist
Threat Intelligence > Blocklist lists every blocked IP address and range, with its source:
- Manual: added by an administrator.
- Behavior: blocked automatically when its behaviour score reached the threshold.
- IP Intel: blocked because of its reputation.
- Feed: imported from a threat feed.
Click Block IP to block an IP or CIDR range, for all services or one service, for a set number of minutes (0 = permanent). Unblock removes a block. In a cluster, manual, behaviour and IP-intel blocks apply on every node.
The blocklist
Behaviour scoring and automatic blocking
Every WAF block, rate-limit hit, failed sign-in and (optionally) backend 4xx error adds points to the client IP's behaviour score. Points fade over time. When the score reaches the Auto-Block Threshold, the IP is blocked on all services for the Block Duration.
Settings are under Settings > Behavior Scoring:
| Setting | Default |
|---|---|
| Auto-Block Threshold | 100 |
| Scoring Window (minutes) | 60 |
| Block Duration (minutes) | 30 (0 = permanent) |
| WAF Block Weight | 10 |
| Rate Limit Weight | 3 |
| Auth Failure Weight | 34 |
| Tunnel Auth Failure Weight | 25 |
| HTTP Error Weight | 2 |
With the defaults, 10 WAF blocks or 3 failed sign-ins within the window block an IP. Threat Intelligence > Behavior Scores shows current scores, with buttons to reset a score or block the IP.
To make sure your own networks are never auto-blocked, add them under Settings > Trusted IPs. Trusted addresses are exempt from behaviour scoring and auto-blocking on all services.
Threat feeds
Threat Intelligence > Threat Feeds imports lists of known-bad IPs into the blocklist on a schedule. Feed types:
- CrowdSec: pull the community blocklist, and optionally push your own detections. Needs a CrowdSec API key.
- AbuseIPDB: pull addresses reported with high confidence. Needs an AbuseIPDB API key.
- Plain Text URL: any URL that returns one IP or CIDR per line. Comments starting with
#are ignored.
Feed Library adds well-known public feeds in one click, including IPsum, Blocklist.de, GreenSnow, Binary Defense and Emerging Threats.
For each feed you set the Pull Interval, the Block Duration for imported addresses, and a Max Entries cap. Each node pulls its feeds itself. Deleting a feed removes its addresses from the blocklist.
Threat feeds
IP intelligence
Threat Intelligence > IP Intel connects a reputation service (AltusIris or IPLocate) that scores each visitor IP from 0 to 100. Add the provider with its API key and a block Threshold, and use Test to look up an address.
Then turn it on per service under IP Intelligence:
- Lookup Mode: Synchronous checks before the request continues. Asynchronous lets the request through and blocks the IP on that service if it's flagged.
- Action Override: Block, Log Only or Challenge. Use Provider Default only logs: choose Block to refuse flagged visitors.
- Block Classifications: block by threat type (VPN, proxy, Tor, hosting, known attacker or abuser).
- Countries: block or allow by country (below).
- IP Intel Bypass IPs / CIDRs: addresses that are never checked.
Private addresses are never looked up. If the provider can't be reached, requests are allowed.
Countries
Countries uses the country from the provider's lookup. Choose a mode, then pick countries from the list:
- Block listed countries: visitors from the selected countries are refused.
- Allow only listed countries (NullWard 0.6.101 or later): visitors from every other country are refused, and so are visitors whose country the provider can't determine.
Only the list of the chosen mode is saved; leave it empty to allow all countries. A refused visitor gets the service's Action Override: with Block, the block page says access from their region isn't permitted. With Log Only or Use Provider Default they are only logged, and the service page warns about it. In the API the lists are ipintel_blocked_countries and ipintel_allowed_countries, as ISO country codes such as US (English country names are accepted and stored as codes).
[!WARNING] An allow list isn't a hard boundary: private addresses and the bypass CIDRs are never checked, and requests are allowed while the provider can't be reached. Combine it with authentication for services that must not be reachable from elsewhere.
IP allowlists
An IP allowlist restricts a service to known networks, for example an internal tool that only your office and VPN should reach.
- Create a list under Threat Intelligence > IP Allowlists with Add Allowlist, entering one IP or CIDR per line.
- On the service, under IP Allowlist, select the list.
The allowlist is checked before every other security layer. All other IP addresses get a 403.
To restrict the admin dashboard itself, use Settings > Security > Admin IP Allowlist. See Deploy with Docker Compose.
Bot detection
Bot detection makes visitors pass a CAPTCHA-style challenge before reaching a service.
- Add a provider under Access Control > Bot Providers: Cloudflare Turnstile, Google reCAPTCHA or hCaptcha, with its site key and secret key.
- On the service, under Bot Detection, turn on Bot Detection Enabled and choose the provider.
Visitors without a valid pass see a challenge page. A pass lasts 24 hours. Use Bot Detection Bypass IPs / CIDRs for monitoring systems and other automated clients.
Malware scanning
NullWard can scan uploads with ClamAV before they reach your application.
- Run ClamAV and point NullWard at it under Settings > Malware Scanner (ClamAV Host, ClamAV Port). See Deploy with Docker Compose.
- On the service, under Malware Scanning, turn on Malware Scanning Enabled and choose Block or Log Only. Use Exempt Paths for endpoints that shouldn't be scanned.
For multipart uploads, each file is scanned. Other request bodies are scanned whole. Malware Detection > Malware Scanner shows the ClamAV version, signature date and each service's scanning status.
[!NOTE] If ClamAV can't be reached, requests are let through, so an outage of the scanner doesn't take your applications down. Watch the scanner status page.

