dotnet-security-owasp

Imported from wshaddix/dotnet-skills (skills/dotnet-security-owasp/SKILL.md). Install upstream with npx skills add wshaddix/dotnet-skills --skill dotnet-security-owasp. Copyright stays with the author.

dotnet-security-owasp

OWASP Top 10 (2021) security guidance for .NET applications. Each category includes the vulnerability description, .NET-specific risk, mitigation code examples, and common pitfalls. This skill is the canonical owner of deprecated security pattern warnings (CAS, APTCA, .NET Remoting, DCOM, BinaryFormatter).

Out of scope: Authentication/authorization implementation -- API-level auth patterns (Identity, OAuth/OIDC, JWT, passkeys, CORS) -- see [skill:dotnet-api-security]. Blazor auth UI (AuthorizeView, CascadingAuthenticationState) -- see [skill:dotnet-blazor-auth]. Cloud-specific security services (Azure Key Vault, AWS Secrets Manager) -- cloud epics. Cryptographic algorithm selection and key management -- see [skill:dotnet-cryptography]. Configuration binding and Options pattern -- see [skill:dotnet-csharp-configuration].

Cross-references: [skill:dotnet-secrets-management] for secrets handling, [skill:dotnet-cryptography] for cryptographic best practices, [skill:dotnet-csharp-coding-standards] for secure coding conventions.

A01: Broken Access Control

Vulnerability: Users act outside their intended permissions -- accessing other users' data, elevating privileges, or bypassing access checks.

Risk in .NET: Missing [Authorize] attributes on controllers/endpoints, insecure direct object references (IDOR) where user IDs are taken from route parameters without ownership validation, and CORS misconfiguration allowing unintended origins.

Mitigation

// 1. Apply authorization globally, then opt out explicitly
builder.Services.AddAuthorizationBuilder()
    .SetFallbackPolicy(new AuthorizationPolicyBuilder()
        .RequireAuthenticatedUser()
        .Build());

var app = builder.Build();
app.MapControllers(); // All endpoints require auth by default

// 2. Resource-based authorization to prevent IDOR
public sealed class DocumentAuthorizationHandler
    : AuthorizationHandler<EditRequirement, Document>
{
    protected override Task HandleRequirementAsync(
        AuthorizationHandlerContext context,
        EditRequirement requirement,
        Document resource)
    {
        if (resource.OwnerId == context.User.FindFirstValue(ClaimTypes.NameIdentifier))
        {
            context.Succeed(requirement);
        }
        return Task.CompletedTask;
    }
}

// In the endpoint:
app.MapPut("/documents/{id}", async (
    int id,
    DocumentDto dto,
    IAuthorizationService authService,
    ClaimsPrincipal user,
    AppDbContext db) =>
{
    var document = await db.Documents.FindAsync(id);
    if (document is null) return Results.NotFound();

    var authResult = await authService.AuthorizeAsync(user, document, "Edit");
    if (!authResult.Succeeded) return Results.Forbid();

    document.Title = dto.Title;
    await db.SaveChangesAsync();
    return Results.NoContent();
});

// 3. Restrict CORS to known origins
builder.Services.AddCors(options =>
{
    options.AddPolicy("Strict", policy =>
    {
        policy.WithOrigins("https://app.example.com")
              .WithMethods("GET", "POST")
              .WithHeaders("Content-Type", "Authorization");
    });
});

Gotcha: AllowAnyOrigin() combined with AllowCredentials() is rejected at runtime by ASP.NET Core, but SetIsOriginAllowed(_ => true) with AllowCredentials() silently allows all origins -- never use this pattern.

A02: Cryptographic Failures

Vulnerability: Sensitive data exposed due to weak or missing encryption -- plaintext storage, deprecated algorithms, or improper key management.

Risk in .NET: Using MD5/SHA1 for hashing passwords, storing connection strings with plaintext passwords in appsettings.json, transmitting sensitive data over HTTP, or using DES/RC2 for encryption.

Mitigation

// Enforce HTTPS and HSTS
builder.Services.AddHttpsRedirection(options =>
{
    options.HttpsPort = 443;
});

var app = builder.Build();
app.UseHsts(); // Strict-Transport-Security header
app.UseHttpsRedirection();

// Never store secrets in appsettings.json -- use user secrets or env vars
// See [skill:dotnet-secrets-management] for proper secrets handling

// Use Data Protection API for symmetric encryption of application data
public sealed class TokenProtector(IDataProtectionProvider provider)
{
    private readonly IDataProtector _protector =
        provider.CreateProtector("Tokens.V1");

    public string Protect(string plaintext) => _protector.Protect(plaintext);

    public string Unprotect(string ciphertext) => _protector.Unprotect(ciphertext);
}

See [skill:dotnet-cryptography] for algorithm selection (AES-GCM, RSA, ECDSA) and key derivation.

A03: Injection

Vulnerability: Untrusted data sent to an interpreter as part of a command or query -- SQL injection, command injection, LDAP injection, and cross-site scripting (XSS).

Risk in .NET: String concatenation in SQL queries, Process.Start with unsanitized input, rendering user input as raw HTML in Razor pages.

Mitigation

// SQL injection prevention: always use parameterized queries
// EF Core is parameterized by default via LINQ
var orders = await db.Orders
    .Where(o => o.CustomerId == customerId)
    .ToListAsync();

// When raw SQL is needed, use parameterized interpolation
var results = await db.Orders
    .FromSqlInterpolated($"SELECT * FROM Orders WHERE Status = {status}")
    .ToListAsync();

// NEVER concatenate user input into SQL:
// var bad = db.Orders.FromSqlRaw("SELECT * FROM Orders WHERE Status = '" + status + "'");

// XSS prevention: Razor encodes output by default.
// Use @Html.Raw() ONLY for trusted, pre-sanitized HTML.
// In Minimal APIs, return typed results -- not raw strings:
app.MapGet("/greeting", (string name) =>
    Results.Content($"<p>Hello, {HtmlEncoder.Default.Encode(name)}</p>",
        "text/html"));

// Command injection prevention: avoid Process.Start with user input.
// If unavoidable, validate against an allowlist:
public static bool IsAllowedTool(string toolName) =>
    toolName is "dotnet" or "git" or "nuget";

Gotcha: FromSqlRaw with string concatenation bypasses parameterization. Always use FromSqlInterpolated or pass SqlParameter objects to FromSqlRaw.

A04: Insecure Design

Vulnerability: Flaws in design patterns that cannot be fixed by implementation alone -- missing rate limiting, lack of defense in depth, unrestricted resource consumption.

Risk in .NET: APIs without rate limiting, unbounded file uploads, missing anti-forgery tokens on state-changing operations.

Mitigation

// Rate limiting with built-in middleware (.NET 7+)
builder.Services.AddRateLimiter(options =>
{
    options.AddFixedWindowLimiter("api", limiterOptions =>
    {
        limiterOptions.PermitLimit = 100;
        limiterOptions.Window = TimeSpan.FromMinutes(1);
        limiterOptions.QueueLimit = 0;
    });
    options.RejectionStatusCode = StatusCodes.Status429TooManyRequests;
});

var app = builder.Build();
app.UseRateLimiter();

app.MapGet("/api/data", () => Results.Ok("data"))
    .RequireRateLimiting("api");

// Anti-forgery for Minimal APIs (.NET 8+)
builder.Services.AddAntiforgery();

var app = builder.Build();
app.UseAntiforgery();

// Form-bound endpoint: antiforgery validated automatically
app.MapPost("/orders", async ([FromForm] string productId, AppDbContext db) =>
{
    var order = new Order { ProductId = productId };
    db.Orders.Add(order);
    await db.SaveChangesAsync();
    return Results.Created($"/orders/{order.Id}", order);
});

// JSON endpoint: opt in explicitly with RequireAntiforgery()
app.MapPost("/api/orders", async (CreateOrderDto dto, AppDbContext db) =>
{
    var order = new Order { ProductId = dto.ProductId };
    db.Orders.Add(order);
    await db.SaveChangesAsync();
    return Results.Created($"/api/orders/{order.Id}", order);
}).RequireAntiforgery();

Gotcha: UseRateLimiter() must be called after UseRouting() and before MapControllers()/MapGet() to apply correctly.

A05: Security Misconfiguration

Vulnerability: Insecure default configurations, incomplete configurations, open cloud storage, unnecessary features enabled, verbose error messages.

Risk in .NET: Detailed exception pages in production (UseDeveloperExceptionPage), default Kestrel settings exposing server headers, debug endpoints left enabled, or missing security headers.

Mitigation

// Remove server identity headers (configure BEFORE Build)
builder.WebHost.ConfigureKestrel(options =>
{
    options.AddServerHeader = false;
});

var app = builder.Build();

if (app.Environment.IsDevelopment())
{
    app.UseDeveloperExceptionPage();
}
else
{
    // Generic error handler in production -- no stack traces
    app.UseExceptionHandler("/error");
    app.UseHsts();
}

// Add security headers via middleware
app.Use(async (context, next) =>
{
    context.Response.Headers.Append("X-Content-Type-Options", "nosniff");
    context.Response.Headers.Append("X-Frame-Options", "DENY");
    context.Response.Headers.Append("Referrer-Policy", "strict-origin-when-cross-origin");
    context.Response.Headers.Append(
        "Content-Security-Policy",
        "default-src 'self'; script-src 'self'; style-src 'self'");
    await next();
});

// Constrain request body size to prevent resource exhaustion (configure BEFORE Build)
builder.WebHost.ConfigureKestrel(options =>
{
    options.Limits.MaxRequestBodySize = 10 * 1024 * 1024; // 10 MB
    options.Limits.MaxRequestHeadersTotalSize = 32 * 1024; // 32 KB
    options.Limits.RequestHeadersTimeout = TimeSpan.FromSeconds(30);
});

// Then: var app = builder.Build();

Gotcha: UseDeveloperExceptionPage() leaks source code paths and stack traces. Ensure it is gated behind IsDevelopment() and never enabled in production or staging.

A06: Vulnerable and Outdated Components

Vulnerability: Using components with known vulnerabilities, unsupported frameworks, or unpatched dependencies.

Risk in .NET: Running on out-of-support .NET versions, NuGet packages with known CVEs, transitive dependency vulnerabilities not audited.

Mitigation

<!-- Enable NuGet audit in Directory.Build.props or csproj -->
<PropertyGroup>
  <NuGetAudit>true</NuGetAudit>
  <NuGetAuditLevel>low</NuGetAuditLevel>
  <NuGetAuditMode>all</NuGetAuditMode> <!-- Audit direct + transitive -->
</PropertyGroup>

# Audit NuGet packages for known vulnerabilities
dotnet list package --vulnerable --include-transitive

# Keep packages up to date
dotnet outdated  # requires dotnet-outdated-tool

# Check .NET SDK/runtime support status
dotnet --info

Gotcha: NuGetAuditMode defaults to direct -- transitive vulnerabilities are hidden unless you set all. Always use all in CI to catch deep dependency issues.

A07: Identification and Authentication Failures

Vulnerability: Weak authentication mechanisms, credential stuffing, session fixation, missing multi-factor authentication.

Risk in .NET: Default Identity password policies that are too weak, session cookies without Secure/SameSite attributes, missing account lockout configuration.

Mitigation

// Configure strong Identity password and lockout policies
builder.Services.AddIdentity<ApplicationUser, IdentityRole>(options =>
{
    // Password requirements
    options.Password.RequireDigit = true;
    options.Password.RequiredLength = 12;
    options.Password.RequireNonAlphanumeric = true;
    options.Password.RequireUppercase = true;
    options.Password.RequireLowercase = true;

    // Account lockout
    options.Lockout.DefaultLockoutTimeSpan = TimeSpan.FromMinutes(15);
    options.Lockout.MaxFailedAccessAttempts = 5;
    options.Lockout.AllowedForNewUsers = true;

    // User settings
    options.User.RequireUniqueEmail = true;
})
.AddEntityFrameworkStores<AppDbContext>()
.AddDefaultTokenProviders();

// Secure cookie configuration
builder.Services.ConfigureApplicationCookie(options =>
{
    options.Cookie.HttpOnly = true;
    options.Cookie.SecurePolicy = CookieSecurePolicy.Always;
    options.Cookie.SameSite = SameSiteMode.Strict;
    options.ExpireTimeSpan = TimeSpan.FromHours(2);
    options.SlidingExpiration = true;
});

Gotcha: CookieSecurePolicy.SameAsRequest allows cookies over HTTP in development, which is fine. But in production behind a reverse proxy terminating TLS, the app sees HTTP -- so cookies are sent insecurely. Always use CookieSecurePolicy.Always in production and configure forwarded headers.

A08: Software and Data Integrity Failures

Vulnerability: Code and infrastructure that does not protect against integrity violations -- unsigned packages, insecure CI/CD pipelines, deserialization of untrusted data.

Risk in .NET: Using BinaryFormatter for deserialization (arbitrary code execution), accepting unsigned NuGet packages from untrusted feeds, missing package source mapping.

Mitigation

<!-- NuGet package source mapping in nuget.config -->
<!-- Only allow packages from trusted sources -->
<configuration>
  <packageSources>
    <add key="nuget.org" value="https://api.nuget.org/v3/index.json" />
    <add key="internal" value="https://pkgs.example.com/nuget/v3/index.json" />
  </packageSources>
  <packageSourceMapping>
    <packageSource key="nuget.org">
      <package pattern="*" />
    </packageSource>
    <packageSource key="internal">
      <package pattern="MyCompany.*" />
    </packageSource>
  </packageSourceMapping>
</configuration>

// NEVER use BinaryFormatter -- it is a critical deserialization vulnerability.
// BinaryFormatter is obsolete as error (SYSLIB0011) in .NET 8+ and removed in .NET 9+.
// Use System.Text.Json instead:
var data = JsonSerializer.Deserialize<OrderDto>(jsonString);

// For binary serialization needs, use MessagePack or Protobuf:
// <PackageReference Include="MessagePack" Version="3.*" />
var bytes = MessagePackSerializer.Serialize(order);
var restored = MessagePackSerializer.Deserialize<Order>(bytes);

Gotcha: Package source mapping uses most-specific-pattern-wins: MyCompany.* beats the * wildcard. Always define specific patterns for internal packages to prevent dependency confusion attacks.

A09: Security Logging and Monitoring Failures

Vulnerability: Insufficient logging of security-relevant events, lack of monitoring for breaches, inability to detect and respond to active attacks.

Risk in .NET: Not logging authentication failures, missing audit trails for sensitive operations, logging sensitive data (passwords, tokens) in plaintext.

Mitigation

// Log security events with structured logging
public sealed class AuditMiddleware(RequestDelegate next, ILogger<AuditMiddleware> logger)
{
    public async Task InvokeAsync(HttpContext context)
    {
        var userId = context.User.FindFirstValue(ClaimTypes.NameIdentifier) ?? "anonymous";
        var path = context.Request.Path.Value;

        using (logger.BeginScope(new Dictionary<string, object?>
        {
            ["UserId"] = userId,
            ["RequestPath"] = path,
            ["RemoteIp"] = context.Connection.RemoteIpAddress?.ToString()
        }))
        {
            await next(context);

            // Log failed authentication attempts
            if (context.Response.StatusCode == StatusCodes.Status401Unauthorized)
            {
                logger.LogWarning("Authentication failed for {Path}", path);
            }

            // Log authorization failures
            if (context.Response.StatusCode == StatusCodes.Status403Forbidden)
            {
                logger.LogWarning("Authorization denied for {Path}", path);
            }
        }
    }
}

// NEVER log sensitive data -- redact credentials and PII
// Configure log filtering to exclude sensitive paths
builder.Logging.AddFilter("Microsoft.AspNetCore.Authentication", LogLevel.Warning);

// Use IHttpLoggingInterceptor (.NET 8+) to redact request/response headers
builder.Services.AddHttpLogging(options =>
{
    options.LoggingFields = HttpLoggingFields.RequestPath
        | HttpLoggingFields.RequestMethod
        | HttpLoggingFields.ResponseStatusCode
        | HttpLoggingFields.Duration;
    // Explicitly exclude request/response bodies and auth headers
});

Gotcha: Structured logging with {Placeholder} syntax is safe, but string interpolation ($"User {userId}") in log calls bypasses structured logging and may leak PII into log sinks that do not support redaction.

A10: Server-Side Request Forgery (SSRF)

Vulnerability: Application fetches a remote resource based on user-supplied URL without validation, allowing attackers to reach internal services or metadata endpoints.

Risk in .NET: HttpClient calls with user-provided URLs, URL redirect following to internal networks, accessing cloud metadata endpoints (169.254.169.254).

Mitigation

// Validate and restrict outbound URLs
public static class UrlValidator
{
    private static readonly HashSet<string> AllowedHosts = new(StringComparer.OrdinalIgnoreCase)
    {
        "api.example.com",
        "cdn.example.com"
    };

    public static bool IsAllowed(string url)
    {
        if (!Uri.TryCreate(url, UriKind.Absolute, out var uri))
            return false;

        // Block non-HTTPS
        if (uri.Scheme != Uri.UriSchemeHttps)
            return false;

        // Block private/internal IPs
        if (IPAddress.TryParse(uri.Host, out var ip))
        {
            if (IsPrivateOrReserved(ip))
                return false;
        }

        // Allowlist hosts
        return AllowedHosts.Contains(uri.Host);
    }

    private static bool IsPrivateOrReserved(IPAddress ip)
    {
        byte[] bytes = ip.GetAddressBytes();
        return bytes[0] switch
        {
            10 => true,                                         // 10.0.0.0/8
            127 => true,                                        // 127.0.0.0/8
            169 when bytes[1] == 254 => true,                   // 169.254.0.0/16 (link-local / cloud metadata)
            172 when bytes[1] >= 16 && bytes[1] <= 31 => true,  // 172.16.0.0/12
            192 when bytes[1] == 168 => true,                   // 192.168.0.0/16
            _ => false
        };
    }
}

// Usage in an endpoint
app.MapPost("/fetch", async (FetchRequest request, IHttpClientFactory factory) =>
{
    if (!UrlValidator.IsAllowed(request.Url))
        return Results.BadRequest("URL not allowed");

    var client = factory.CreateClient();
    var response = await client.GetStringAsync(request.Url);
    return Results.Ok(response);
});

// Configure HttpClient to disable automatic redirect following
builder.Services.AddHttpClient("external", client =>
{
    client.BaseAddress = new Uri("https://api.example.com");
})
.ConfigurePrimaryHttpMessageHandler(() => new SocketsHttpHandler
{
    AllowAutoRedirect = false // Prevent redirect-based SSRF
});

Gotcha: DNS rebinding can bypass IP allowlists -- an attacker's domain resolves to a public IP during validation but to an internal IP during the actual request. Pin DNS resolution or re-validate after connection.

Deprecated Security Patterns

This skill is the canonical owner of deprecated security pattern warnings. Other skills should cross-reference here rather than duplicating these warnings.

Code Access Security (CAS)

CAS is not supported in .NET Core/.NET 5+. Code that references System.Security.Permissions, SecurityPermission, or [SecurityCritical]/[SecuritySafeCritical] attributes for CAS purposes must be removed or replaced with OS-level security boundaries (containers, process isolation).

AllowPartiallyTrustedCallers (APTCA)

The [AllowPartiallyTrustedCallers] attribute has no effect in .NET Core/.NET 5+. The partial-trust model is gone. Remove APTCA attributes during migration. Use standard authorization and input validation instead.

.NET Remoting

.NET Remoting is not available in .NET Core/.NET 5+. It was inherently insecure due to unrestricted deserialization of remote objects. Replace with:

• gRPC for cross-process/cross-machine RPC (see [skill:dotnet-cryptography] for transport security)
• Named pipes for same-machine IPC
• HTTP APIs for service-to-service communication

DCOM

Distributed COM (DCOM) is Windows-only and not supported in .NET Core/.NET 5+. Replace with gRPC, REST APIs, or message queues for distributed communication.

BinaryFormatter

BinaryFormatter is obsolete as error (SYSLIB0011) in .NET 8 and removed in .NET 9+. It enables arbitrary code execution through deserialization attacks. Replace with:

• System.Text.Json for JSON serialization
• MessagePack or Protocol Buffers for binary formats
• XmlSerializer with strict type allowlists for XML scenarios

Do not set System.Runtime.Serialization.EnableUnsafeBinaryFormatterSerialization to true as a workaround.

Agent Gotchas

1. Do not use [AllowAnonymous] without explicit justification -- it overrides the global fallback policy. Mark each anonymous endpoint with a comment explaining why.
2. Do not disable HTTPS redirection for convenience -- use dotnet dev-certs https --trust for local development instead.
3. Do not log raw request bodies -- they may contain credentials, tokens, or PII. Use HttpLoggingFields to select safe fields.
4. Do not rely solely on client-side validation -- always validate on the server. Razor form validation is for UX, not security.
5. Do not use FromSqlRaw with string interpolation -- use FromSqlInterpolated which auto-parameterizes.
6. Do not store secrets in appsettings.json -- use user secrets for development and environment variables or managed identity for production. See [skill:dotnet-secrets-management].
7. Do not generate security-sensitive code using deprecated patterns -- CAS, APTCA, .NET Remoting, DCOM, and BinaryFormatter are all unsupported in modern .NET. See the Deprecated Security Patterns section above.

Prerequisites

• .NET 8.0+ (LTS baseline)
• ASP.NET Core 8.0+ for security middleware, anti-forgery, and rate limiting
• Microsoft.AspNetCore.Identity for authentication/identity (if using A07 patterns)

References

• OWASP Top 10 (2021)
• ASP.NET Core Security
• Secure Coding Guidelines for .NET
• Security in .NET
• ASP.NET Core Data Protection
• Rate Limiting Middleware
• NuGet Package Source Mapping
• BinaryFormatter Migration Guide