agentleFS
PunGrumpyPunGrumpy/docker-doctor/skills/docker-author/SKILL.md

docker-author

Skill4 starsChanged 26 days agoMITView on GitHub

Use when writing, scaffolding, or refactoring a Dockerfile or Docker Compose file — "write me a Dockerfile", "containerize/dockerize this app", "add Docker", or editing an existing Dockerfile/compose.yaml. Applies production defaults (multi-stage, non-root, pinned slim base, layer caching, .dockerignore, healthcheck) and self-verifies with docker-doctor. Companion to the `docker-doctor` scan skill.

  • Reads credentials
  • Deletes or force-pushes
  • Installs packages

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write Dockerfiles and Compose files that pass a docker-doctor scan on the first try.

    line 9

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Docker Author

Write Dockerfiles and Compose files that pass a docker-doctor scan on the first try. Every default below maps to a rule the scanner enforces — this skill front-loads them so the generated file starts at a high health score instead of being fixed afterward.

The loop: author → npx @docker-doctor/cli@latest . --verbose → fix anything flagged → aim for Excellent 🏆 (score ≥ 90). Don't hand back a Dockerfile you haven't scanned.

Production defaults (each maps to a docker-doctor rule)

Apply these when authoring. If you deviate, say why.

DefaultRule it satisfies
Multi-stage build — a build/deps stage separate from the slim runtime stageuse-multi-stage, avoid-dev-dependencies
Pin the base image to a concrete tag (never latest, never bare) — prefer digest for reproducibilitypin-image-version
Slim/alpine/distroless runtime base (e.g. node:22-alpine, python:3.12-slim)prefer-slim-base
Run as non-root — create a user and USER node / USER 1000 before CMDno-root-user, useradd-no-log-init
No secrets in ENV/ARG — pass at runtime or use RUN --mount=type=secretno-secrets-in-env
COPY not ADD (except local tar auto-extract); never ADD a remote URLprefer-copy-over-add, no-add-remote
Cache-friendly layer order — copy manifests + install deps before copying sourceorder-layers
One RUN per logical step, &&-chained; apt-get update && apt-get install in the same layerminimize-layers, combine-apt-update-install, avoid-run-cd
Clean package caches in the same layer (rm -rf /var/lib/apt/lists/*, --no-cache, npm ci --omit=dev)clean-package-cache
SHELL ["/bin/bash", "-o", "pipefail", "-c"] before piped RUNsuse-pipefail
Absolute WORKDIR (e.g. /app), set onceabsolute-workdir
Exec-form CMD/ENTRYPOINT — JSON array, so signals propagateuse-exec-form
HEALTHCHECK for long-running servicesrequire-healthcheck
OCI LABELs (org.opencontainers.image.source, .title, …)require-labels
Sorted multi-line args (alphabetize package lists)sort-multiline-args
Ship a .dockerignore (.git, node_modules, build output, secrets)use-dockerignore

Beyond the rules: enable BuildKit and use RUN --mount=type=cache for package managers (faster rebuilds without bloating layers) and RUN --mount=type=secret for build-time credentials (keeps them out of the image).

Canonical multi-stage template (Node — adapt per stack)

# syntax=docker/dockerfile:1
FROM node:22-alpine AS deps
WORKDIR /app
COPY package.json package-lock.json ./
RUN --mount=type=cache,target=/root/.npm npm ci

FROM node:22-alpine AS build
WORKDIR /app
COPY --from=deps /app/node_modules ./node_modules
COPY . .
RUN npm run build

FROM node:22-alpine AS runtime
WORKDIR /app
ENV NODE_ENV=production
LABEL org.opencontainers.image.source="https://github.com/acme/app"
COPY --from=build /app/dist ./dist
COPY --from=deps /app/node_modules ./node_modules
USER node
HEALTHCHECK --interval=30s --timeout=3s CMD wget -q --spider http://localhost:3000/health || exit 1
CMD ["node", "dist/server.js"]

Compose defaults (each maps to a Compose rule)

DefaultRule
No top-level version: key (obsolete in Compose v2)no-version-key
Resource limits — deploy.resources.limits (cpus/memory) per servicerequire-resource-limits
restart: policy (unless-stopped / on-failure)require-restart-policy
depends_on with condition: service_healthy, not the short list formuse-depends-on-condition
Narrow bind mounts — mount ./data, never /, ~, ~/.ssh, ../ or /etcno-broad-bind-mount
Read-only host mounts — :ro / read_only: true on any bind mount outside the project unless the service writes to itprefer-read-only-bind-mount
No Docker socket bind mount — a filtering socket proxy insteadno-docker-socket-mount
No host namespaces — no pid, ipc, uts, cgroup or userns_mode set to host, and shm_size for shared memoryno-host-namespaces
Keep confinement — no SYS_ADMIN, SYS_MODULE, SYS_RAWIO or DAC_READ_SEARCH in cap_add, no unconfined or label:disable in security_opt, and never cap_add: [ALL]no-weakened-confinement, no-privileged-service
Loopback for unauthenticated AI services — bind Ollama, Qdrant and the MCP gateway to 127.0.0.1 or an internal networkno-exposed-unauthenticated-service

Verify before handing off

npx @docker-doctor/cli@latest . --verbose

Fix every ERROR, then WARN. If a flagged rule genuinely doesn't apply, don't silently ignore it — either explain the trade-off or configure it (see the docker-doctor skill's references/explain.md). Inside this repo, run bun packages/docker-doctor/src/cli.ts . instead of npx.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.