Dockerfile Hardening Audit

Statically audit Dockerfiles for common container hardening risks (root user, unpinned/latest base images, missing healthchecks, and risky build patterns).

Security Scan

The skill's code and runtime instructions are consistent with its stated purpose (static Dockerfile hardening checks), request only bash/python3, perform local static analysis of files matched by a configurable glob, and do not contact external endpoints or request secrets.

Purpose & Capability

The name/description (Dockerfile hardening audit) matches the included script and SKILL.md. The required binaries (bash, python3) are exactly what's needed to run the provided shell+Python script. No unrelated credentials, config paths, or binaries are requested.

Instruction Scope

The SKILL.md instructs the agent to run the bundled script which performs static text analysis of Dockerfile paths matched by DOCKERFILE_GLOB. The script reads files from disk only (no network calls), uses regex checks to flag patterns (user, FROM tags, HEALTHCHECK, ADD, remote script pipes), and does not execute code from the Dockerfiles. Inputs are explicit and limited to the documented env-vars.

Install Mechanism

There is no install spec or remote download; the script is bundled in the skill. This is low risk because nothing is fetched from external URLs or written to unusual locations during install.

Credentials

No secrets or external service credentials are requested. The skill accepts a number of optional environment flags to control scanning behavior (glob, thresholds, toggles) which are proportionate to the task.

Persistence & Privilege

always is false and the skill does not modify agent/system configuration or other skills. It runs on-demand and does not require persistent presence or elevated privileges.

Assessment

This skill appears coherent and low-risk for its stated purpose. Before running: (1) review/limit DOCKERFILE_GLOB (or set FILE_MATCH/FILE_EXCLUDE) so you only scan intended repositories/paths, (2) remember it is a static scanner — it looks for text patterns and can produce false positives/negatives (especially for complex multi-stage FROM lines or obfuscated remote fetches), and (3) you can safely inspect the bundled script (skills/dockerfile-hardening-audit/scripts/dockerfile-hardening-audit.sh) yourself to confirm no unexpected behavior. No network calls or secret exfiltration were found.

Use this skill to statically audit Dockerfiles before insecure container defaults land in production.

What this skill does

• Scans Dockerfiles and scores hardening risk per file
• Flags missing non-root USER declarations
• Flags base images using floating tags (:latest, :main, :master, :edge) or no tag/digest
• Flags missing HEALTHCHECK
• Flags ADD instructions (when COPY is safer/clearer)
• Flags curl|bash/wget|sh style remote script execution
• Supports include/exclude regex filters and fail-gate mode

Inputs

Optional:

• DOCKERFILE_GLOB (default: **/Dockerfile*)
• TOP_N (default: 20)
• OUTPUT_FORMAT (text or json, default: text)
• WARN_SCORE (default: 3)
• CRITICAL_SCORE (default: 6)
• REQUIRE_NON_ROOT_USER (0/1, default: 1)
• REQUIRE_HEALTHCHECK (0/1, default: 1)
• FLAG_FLOATING_TAGS (0/1, default: 1)
• FLAG_UNPINNED_IMAGES (0/1, default: 1)
• FLAG_ADD_INSTRUCTIONS (0/1, default: 1)
• FLAG_REMOTE_SCRIPT_PIPE (0/1, default: 1)
• FILE_MATCH (regex include filter on Dockerfile path, optional)
• FILE_EXCLUDE (regex exclude filter on Dockerfile path, optional)
• FAIL_ON_CRITICAL (0 or 1, default: 0)

Run

Text report:

DOCKERFILE_GLOB='**/Dockerfile*' \
bash skills/dockerfile-hardening-audit/scripts/dockerfile-hardening-audit.sh

JSON output + fail gate:

DOCKERFILE_GLOB='**/Dockerfile*' \
OUTPUT_FORMAT=json \
FAIL_ON_CRITICAL=1 \
bash skills/dockerfile-hardening-audit/scripts/dockerfile-hardening-audit.sh

Run against bundled fixtures:

DOCKERFILE_GLOB='skills/dockerfile-hardening-audit/fixtures/*Dockerfile*' \
bash skills/dockerfile-hardening-audit/scripts/dockerfile-hardening-audit.sh

Output contract

• Exit 0 in report mode (default)
• Exit 1 when FAIL_ON_CRITICAL=1 and one or more Dockerfiles are critical
• Text mode prints summary + ranked Dockerfile risks
• JSON mode prints summary + ranked Dockerfiles + critical Dockerfiles