Zetian.AntiSpam

Advanced Spam Protection for Zetian SMTP Server

📦 NuGet Package ⭐ Star on GitHub

Bayesian Filtering

Machine learning-based spam detection

SPF/DKIM/DMARC

Complete email authentication suite

RBL/DNSBL

Realtime blackhole list checking

Greylisting

Temporary rejection for unknown senders

Custom Filters

Extensible with custom spam checkers

Statistics

Detailed metrics and reporting

Quick Start

Basic Anti-Spam Setup

using Zetian.Server;
using Zetian.AntiSpam.Extensions;

// Create SMTP server
var server = new SmtpServerBuilder()
    .Port(25)
    .ServerName("My SMTP Server")
    .Build();

// Add anti-spam with default settings
server.AddAntiSpam();

await server.StartAsync();

Custom Configuration

// Configure anti-spam features
server.AddAntiSpam(builder => builder
    .EnableSpf(failScore: 50)
    .EnableRbl("zen.spamhaus.org", "bl.spamcop.net")
    .EnableBayesian(spamThreshold: 0.85)
    .EnableGreylisting(initialDelay: TimeSpan.FromMinutes(5))
    .WithOptions(options =>
    {
        options.RejectThreshold = 60;
        options.TempFailThreshold = 40;
        options.RunChecksInParallel = true;
    }));

Lenient Protection

// For trusted environments
server.AddAntiSpam(builder => 
    builder.UseLenient());

Aggressive Protection

// Maximum spam protection
server.AddAntiSpam(builder => 
    builder.UseAggressive());

Email Authentication (SPF, DKIM, DMARC)

SPF (Sender Policy Framework)

Verifies that the sending server is authorized to send email for the domain:

// Add SPF checking
server.AddSpfCheck(failScore: 50);

SPF Results & Scores

Pass0 points

None5 points

Neutral10 points

SoftFail30 points

Fail50 points

DKIM (DomainKeys Identified Mail)

Verifies email authenticity using digital signatures:

// Add DKIM checking with strict mode
server.AddDkimCheck(
    failScore: 40,
    strictMode: true  // Require all essential headers to be signed
);

DMARC (Domain-based Message Authentication)

Enforces policies based on SPF and DKIM results:

// Add DMARC checking with policy enforcement
server.AddDmarcCheck(
    failScore: 70,        // Score for reject policy
    quarantineScore: 50,  // Score for quarantine policy
    enforcePolicy: true   // Enforce DMARC policies
);

Complete Email Authentication

// Add all three authentication methods in one call
server.AddEmailAuthentication(
    strictMode: true,      // Strict DKIM validation
    enforcePolicy: true    // Enforce DMARC policies
);

RBL/DNSBL Checking

// Add RBL checking with popular lists
server.AddRblCheck(
    "zen.spamhaus.org",       // Spamhaus ZEN
    "bl.spamcop.net",         // SpamCop
    "b.barracudacentral.org"  // Barracuda
);

Bayesian Filtering

Machine Learning Spam Detection

// Option 1: Add Bayesian to server
server.AddBayesianFilter(spamThreshold: 0.9);

// Option 2: Work with Bayesian directly for training
var bayesianFilter = new BayesianSpamFilter(spamThreshold: 0.9);
await bayesianFilter.TrainSpamAsync("spam content");
await bayesianFilter.TrainHamAsync("legitimate content");

// Then add trained filter to server
server.AddSpamChecker(bayesianFilter);

Training the Filter

// Create and train Bayesian filter
var bayesianFilter = new BayesianSpamFilter();

// Load training data from files
var spamEmails = Directory.GetFiles("spam/", "*.eml")
    .Select(File.ReadAllText);

var hamEmails = Directory.GetFiles("ham/", "*.eml")
    .Select(File.ReadAllText);

// Train the filter
foreach (var spam in spamEmails)
{
    await bayesianFilter.TrainSpamAsync(spam);
}

foreach (var ham in hamEmails)
{
    await bayesianFilter.TrainHamAsync(ham);
}

// Add trained filter to server
server.AddSpamChecker(bayesianFilter);

Greylisting

// Add greylisting
server.AddGreylisting(initialDelay: TimeSpan.FromMinutes(5));

// With domain whitelisting
var greylistingChecker = new GreylistingChecker(
    initialDelay: TimeSpan.FromMinutes(5));
greylistingChecker.Whitelist("trusted.com");
server.AddSpamChecker(greylistingChecker);

How It Works

1. First email attempt is temporarily rejected
2. Legitimate servers retry after a delay
3. Retry is accepted and sender is whitelisted
4. Spam bots rarely retry, filtering them out

Custom Spam Checker

using Zetian.AntiSpam.Abstractions;

public class CustomSpamChecker : ISpamChecker
{
    public string Name => "Custom";
    public bool IsEnabled { get; set; } = true;

    public async Task<SpamCheckResult> CheckAsync(
        ISmtpMessage message,
        ISmtpSession session,
        CancellationToken cancellationToken)
    {
        // Custom logic here
        if (IsSpam(message))
        {
            return SpamCheckResult.Spam(75, "Custom rule triggered");
        }
        
        return SpamCheckResult.Clean(0);
    }
    
    private bool IsSpam(ISmtpMessage message)
    {
        // Check for suspicious patterns
        var subject = message.Subject?.ToLower() ?? "";
        
        // Example: Check for common spam keywords
        var spamKeywords = new[] { "viagra", "lottery", "winner", "prize" };
        if (spamKeywords.Any(keyword => subject.Contains(keyword)))
            return true;
            
        // Example: Check for excessive caps
        if (subject.Count(char.IsUpper) > subject.Length * 0.8)
            return true;
            
        return false;
    }
}

// Add custom checker
server.AddSpamChecker(new CustomSpamChecker());

Score Interpretation

Score	Action	Description
0-30	Accept	Clean message
30-50	Accept with caution	Possible spam
50-70	Greylist/Temp Reject	Likely spam
70-90	Reject	High confidence spam
90-100	Hard Reject	Definite spam

Best Practices

Start with lenient settings and adjust
Train Bayesian filters with balanced data
Whitelist trusted partners

Monitor block rates regularly
Combine multiple detection methods
Keep RBL lists updated

← Relay Extension Back to Extensions →