Skip to main content

Spam Filtering

Spam Filtering

WildDuck integrates with Rspamd for spam detection. Rspamd is a fast, free spam filtering system that uses various techniques including content analysis, DNS blocklists, SPF/DKIM verification, and machine learning.

How It Works​

Rspamd is integrated through the Haraka inbound SMTP pipeline:

  1. Haraka receives a message on the SMTP port
  2. haraka-plugin-rspamd sends the message to Rspamd for scanning
  3. Rspamd analyzes the message and returns a spam score and action recommendation
  4. haraka-plugin-wildduck reads the Rspamd results and routes the message:
    • Messages flagged as spam → Junk folder
    • Clean messages → INBOX (or the folder specified by user filters)

Rspamd Scoring​

Rspamd assigns a numerical score to each message based on multiple factors:

  • Content analysis (spam patterns, phishing detection)
  • DNS blocklists (DNSBL/RBL checks)
  • SPF, DKIM, and DMARC verification results
  • Fuzzy hash matching against known spam
  • Bayesian classifier (learns from user actions)
  • URL reputation

The haraka-plugin-wildduck uses Rspamd's action recommendations (reject, greylist, add header, no action) to determine how to handle the message.

Setup​

To use Rspamd with WildDuck:

  1. Install Rspamd on your server (included in the WildDuck setup script)
  2. Enable the haraka-plugin-rspamd in your Haraka configuration
  3. Ensure haraka-plugin-wildduck is configured to read Rspamd results

The WildDuck scripted install and Docker setup both include Rspamd pre-configured.