A secret written into ENV or ARG stays in every image layer and in the build history, so anyone who can pull the image can read it. This skill finds that and nine other common Dockerfile mistakes, and tells you how to fix each one.

Your agent runs a scanner on your Dockerfiles (Dockerfile, Dockerfile.*, *.dockerfile, Containerfile). It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach for every finding. Multi-stage builds are understood: the root user and health checks are judged on the final stage. Secret values are never printed: the report shows the variable name and length only.

What it checks

BLOCK (fix before shipping)

• D01: a password, token or key written into ENV or ARG

• D02: a script piped from curl or wget straight into a shell

• D03: a .env file, an SSH or TLS private key, or a .p12/.pfx bundle copied into the image

REVIEW (check before release)

• D11: the final image runs as root

• D12: a base image with no tag or :latest

• D13: ADD of a remote URL with no checksum

• D14: apt-get install without removing the package lists

• D15: an exposed port with no HEALTHCHECK

• D16: sudo in a build step

• D17: chmod 777

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 Dockerfile, and the real report the scanner produces on the risky one

Limits

It is a static check of the Dockerfile text. It does not build images or read build arguments, and it supports the default backslash line continuation only. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 7 files: SKILL.md, the scanner in scripts/ (2 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (2 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.