An echo "$API_TOKEN" in a script puts the token in the job log, and a .env file listed under artifacts can be downloaded by anyone who can see the pipeline. This skill finds those and eight other GitLab CI risks, and tells you how to fix each one.

Your agent runs a scanner on .gitlab-ci.yml, *.gitlab-ci.yml files and the YAML under .gitlab/. It reads global settings, default:, hidden jobs and every job, follows YAML anchors and extends, and skips !reference lines. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach. Secret values are never printed.

What it checks

BLOCK (fix before shipping)

• G01: a password, token or key written in variables

• G02: curl or wget piped into a shell

• G03: a secret variable echoed to the job log

• G04: a .env file, private key or Terraform state kept as an artifact

REVIEW (check before release)

• G11: an image or service with no tag or :latest

• G12: an include from a URL, or a project or component not pinned to a tag

• G13: a security scan job with allow_failure: true

• G14: TLS checks turned off (curl -k, GIT_SSL_NO_VERIFY)

• G15: env or printenv dumping every variable to the log

• G16: a production deploy job with no rules, only or manual step

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

Limits

It reads each file on its own and does not fetch included files or see CI/CD variables set in GitLab settings. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 8 files: SKILL.md, the scanner in scripts/ (3 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.