One conn.Query<User>($"... WHERE Email = '{email}'") or one new BinaryFormatter().Deserialize(Request.Body) is enough to lose your data or your server. This skill finds those and eleven other C# security mistakes before you deploy, and tells you how to fix each one.

Your agent runs a scanner on your .cs files and your .cshtml and .razor views. It skips comments, tests, generated code and the bin and obj folders, understands verbatim, interpolated and raw strings including the values inside {...}, follows SQL built in a variable or a StringBuilder to the call that runs it, and knows which values come from the request (controller, Razor Pages and minimal API parameters, [FromQuery], [BindProperty], Request.Query, IFormFile.FileName). It knows the safe patterns, so ordinary EF Core and Dapper code stays quiet: FromSql and FromSqlInterpolated, ExecuteSqlInterpolated, FromSqlRaw with {0} placeholders, @Name parameters, LINQ, constants and int values, sort columns picked by a switch or checked against a list, Path.GetFileName, Url.IsLocalUrl and RandomNumberGenerator. It never compiles or runs C#. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach. Secrets are never printed.

What it checks

BLOCK (fix before shipping)

• C01: SQL built with +, $"...{value}", string.Format or StringBuilder and run with SqlCommand, CommandText, Dapper, EF Core FromSqlRaw, ExecuteSqlRaw or SqlQueryRaw, EF6 SqlQuery or NHibernate

• C02: Process.Start or ProcessStartInfo running cmd, powershell or bash with a changing command, or a program or Arguments string from the request

• C03: BinaryFormatter, SoapFormatter, NetDataContractSerializer, LosFormatter, Json.NET TypeNameHandling without a binder, Type.GetType with a type name from the request

• C04: hard-coded passwords, API keys, JWT signing keys, client secrets, credentials, literal passwords in Identity seed code (UserManager.CreateAsync) and Password= or AccountKey= in connection strings

REVIEW (check before release)

• C11: certificate validation callbacks that always return true and DangerousAcceptAnyServerCertificateValidator

• C12: DES, TripleDES, RC2, ECB mode and fixed IVs

• C13: MD5 or SHA on passwords, Rfc2898DeriveBytes or PBKDF2 under 100,000 iterations, Identity V2 hashes

• C14: System.Random for tokens, OTPs, salts and reset codes

• C15: Html.Raw, HtmlString and MarkupString with values that are not fixed text (XSS)

• C16: XmlUrlResolver, DtdProcessing.Parse and trusted XSLT (XXE)

• C17: file paths from the request in Path.Combine, File.*, FileStream or PhysicalFile

• C18: Redirect(returnUrl) without Url.IsLocalUrl (open redirect)

• C19: passwords, tokens, connection strings and the Authorization header written to logs

What's included

• The scanner (Python 3.9+, standard library only)

• A rules guide with the reason and safer approach for each rule

• A risky and a safe example ASP.NET Core shop API with a Razor view, and the real report the scanner produces on the risky one

Limits

It reads code as text, not with the C# compiler: types are not known and values are followed within one file. appsettings.json, web.config, .csproj files, NuGet packages and known CVEs are not checked, and Razor views are checked only for raw HTML output. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 24 files: SKILL.md, the scanner in scripts/ (7 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (14 example files plus report.md, the real scanner output on the risky example).

How to use: This is an agent skill: a folder with a SKILL.md file and Python scripts that an AI coding agent such as Claude Code, Cursor, or Codex CLI can run. Unzip it into your agent's skills folder (for Claude Code: ~/.claude/skills/), then ask your agent to check your project. The scripts also run on their own with Python 3.9 or newer, with nothing to install. They read your files only: no network access, and nothing is sent anywhere.