agentleFS
PaldomPaldom/mac-os-setup-skills/skills/docker-on-mac/SKILL.md

docker-on-mac

Skill0 starsChanged 3 months agoMITView on GitHub

Sets up a container runtime on macOS - Docker Desktop (paid over 250 staff/$10M), OrbStack (paid commercial use), or Colima (free CLI) - install, hello-world verify, contexts, amd64 on Apple Silicon. Use for "install docker on my Mac", "docker desktop vs orbstack vs colima", "docker command not found", "exec format error". Not for Dockerfiles, compose debugging, Kubernetes, or databases.

  • Installs packages

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    Never uninstall an existing runtime without consent.

    line 50
  2. Distinctive:

    Make sure you're testing the runtime you just installed, not a leftover one — select its context first:

    line 77
  3. Very specific:

    Default build/run platform is linux/arm64.

    line 90
  4. Distinctive:

    Prefer native arm64 images locally.

    line 103

≈ “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 on Mac

Chooses and installs the right container runtime for the user's licensing situation and workflow. macOS runs Linux containers in a VM — the runtime choice is really a choice of VM manager, and in 2026 it's a licensing decision as much as a technical one.

When NOT to use

  • Writing Dockerfiles / compose files, image debugging → native competence
  • Kubernetes / cloud contexts → cloud-dev-setup
  • Local Postgres/Redis without containers → web-dev-setup
  • Windows/Linux desktop VMs → Parallels/UTM via mac-dev-apps

Prerequisites

brew present (→ homebrew-setup).

The decision (ask these two questions)

  1. Licensing: Docker Desktop is free only for personal use, education, non-commercial OSS, and businesses with fewer than 250 employees AND under $10M revenue — beyond either bound it needs a paid subscription. Any commercial use → OrbStack requires Pro (~$8/user/mo). Colima and the plain Docker CLI are free for everything.
  2. Workflow: want a GUI/dashboard → Desktop or OrbStack. Terminal-only fine → Colima.
Docker DesktopOrbStackColima
Costfree under 250 emp AND $10M; else paidfree personal; Pro for commercialfree, open source
Speed/RAMheaviestfastest start, lowest idlemiddle
GUIyesyesno
Fitcross-platform teams, org standardMac-first daily driverfree-policy orgs, CI-like

(Apple's own container tool is 1.x and macOS-26-only, no compose — watch, don't adopt as the daily driver.)

Safety rails

  • These installs may prompt for the admin password (privileged helper / docker.sock symlink) — expected; tell the user.
  • Never uninstall an existing runtime without consent; multiple runtimes coexist via docker context.

Workflow

Install the chosen runtime

# Docker Desktop  (cask renamed from `docker` in 2025 - old token aliases)
brew install --cask docker-desktop
open -a "Docker"          # first launch: accept terms, grant privileged helper

# OrbStack - bundles docker CLI/compose/buildx if missing
brew install --cask orbstack
open -a OrbStack

# Colima - engine + separate CLI client
brew install colima docker docker-compose
colima start              # Apple Silicon fast path: colima start --vm-type=vz --vz-rosetta

The trap: brew install docker (formula, no --cask) is the CLI client only — no engine. "Installed docker but nothing runs" is almost always this; pair the CLI with a runtime or install a cask.

Verify

Make sure you're testing the runtime you just installed, not a leftover one — select its context first:

docker context ls          # active context marked *; desktop-linux / orbstack / colima
docker context use <the-new-runtime's-context>
docker run --rm hello-world
docker compose version     # compose plugin wired (bundled by Desktop/OrbStack)
sh "${CLAUDE_SKILL_DIR}/scripts/check.sh"

Multi-arch on Apple Silicon

Default build/run platform is linux/arm64. For CI/prod parity:

docker run --platform linux/amd64 <image>        # one-off, emulated (Rosetta) - slower
# Multi-arch image: needs a destination or it only fills the build cache -
# push to a registry (or --load for ONE platform locally):
docker buildx build --platform linux/amd64,linux/arm64 -t registry/you/app:tag --push .
docker buildx imagetools inspect registry/you/app:tag   # confirm both platforms

"exec format error" in CI is often an arm64-only image meeting an amd64 runner (also caused by bad shebangs/CRLF — check the image arch first: docker image inspect --format '{{.Architecture}}' <image>). Prefer native arm64 images locally; treat amd64 emulation as a compatibility path. Rosetta x86 emulation is on by default in current Docker Desktop and OrbStack on Apple Silicon.

Output spec

Done means: chosen runtime installed and started, docker run --rm hello-world succeeds, active context confirmed, licensing constraint recorded in the summary (which tier and why), and the user knows the multi-arch flags if they deploy to amd64.

Gotchas

  • Docker Desktop ↔ OrbStack both want /var/run/docker.sock; the last one granted admin wins the symlink — contexts still route correctly, but tools hardcoding the sock path follow the symlink owner.
  • Colima VMs don't autostart at login by default — brew services start colima if that's wanted.
  • Docker Desktop requires macOS 14+ now; keep it updated via its own updater (auto_updates cask).
  • Bind-mount-heavy workloads (node_modules in containers) are the classic Mac pain — OrbStack's filesystem is the fastest of the three; or keep hot paths in named volumes.
  • License thresholds and prices drift — verify at docker.com/pricing and orbstack.dev/pricing when it matters (references/runtime-licensing.md).

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.