Remove-Item "$Staging\*" -Recurse -Force deletes everything at the root of the drive when $Staging is empty, and one irm https://... | iex runs whatever a server sends, with no file to review and no execution policy check. This skill finds those and ten other PowerShell script mistakes before you run or ship your scripts, and tells you how to fix each one.

Your agent runs a scanner on .ps1 scripts, .psm1 modules and .psd1 data files, including the build/ and dist/ folders where CI scripts often live. It skips comments, knows single-quoted, double-quoted and here-strings and which of them expand $variables, follows pipelines across lines, and knows the usual guards ([Parameter(Mandatory)], IsNullOrWhiteSpace, Join-Path, -ErrorAction Stop). It never runs the script. 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)

• P01: Invoke-Expression (iex) of a variable, an argument or a file read through a variable path

• P02: downloads run as code (irm | iex, iex "$(irm ...)", DownloadString, [scriptblock]::Create, piped into powershell)

• P03: passwords and keys in ConvertTo-SecureString -AsPlainText, variables, .psd1 settings, parameters, Authorization headers and connection strings

• P04: Remove-Item -Recurse on "$dir\*" when dir may be empty, or on C:\

REVIEW (check before release)

• P11: Set-ExecutionPolicy Bypass or Unrestricted beyond the current session

• P12: certificate checks off (-SkipCertificateCheck, ServerCertificateValidationCallback), WinRM TrustedHosts * or curl.exe -k

• P13: commands built from strings (cmd /c, powershell -Command, ScriptBlock::Create)

• P14: plain http downloads and unencrypted authentication

• P15: -EncodedCommand and Base64 text run as code

• P16: secrets printed with Write-Host or traced with Set-PSDebug -Trace 2

• P17: Install-Module without a pinned version, or -SkipPublisherCheck

• P18: Set-Location without a check before Remove-Item -Recurse

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 project (deploy and install scripts and a settings file), and the real report the scanner produces on the risky one

Limits

It reads each file as written: dot-sourced files, imported modules, splatted hashtables and aliases you define yourself are not followed, and PowerShell inside CI YAML, .bat or .cmd files is not read. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 13 files: SKILL.md, the scanner in scripts/ (4 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (6 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.