Skip to main content

Getting started

Let's discover Corgi in less than 10 minutes.

Corgi logo

corgi is how you run your project locally — every day, with one command. You describe your stack once in a corgi-compose.yml, and corgi run brings the whole thing up: repos cloned, databases seeded, .env files written, every service started. It even starts Docker for you, so the infra just happens.

Onboarding comes free: hand a teammate the same file and they go from nothing to a running stack in minutes — no setup call, no "works on my machine". And because corgi never blocks on a prompt and speaks plain JSON, an AI agent or CI job can drive it just like you do.

What corgi does for you​

  • Your repos — clones each service from its Git URL, pulls them all at once, forks them, or runs one on a branch worktree without disturbing your checkout.
  • Your databases — 40+ drivers started in Docker and seeded from a dump or remote DB; native shells with credentials filled in (corgi db shell); LocalStack and Supabase stacks from the same file.
  • Your services — everything starts together, in the right order, with env vars wired between them; Ctrl-C cleans up. Run it detached with corgi run -d.
  • The fiddly bits — corgi doctor preflight, live corgi status -w health, public HTTPS tunnels, saved logs, and desktop crash notifications.
  • Made for AI agents — stable JSON and exit codes, an MCP server (corgi mcp), and a Claude Code plugin that ships tickets as draft PRs (/corgi:stories) and reviews them (/corgi:review).

In your day-to-day​

corgi isn't a one-time setup tool — it's how you run the project, in whatever shape the day calls for:

  • corgi run — the whole stack, one command (and it starts Docker for you).
  • corgi db -u — just the databases, when you run a service from your IDE.
  • corgi run --tier staging --services web — run only the frontend, pointed at staging. Declare env tiers once, switch with a flag.
  • New project? Write a corgi-compose.yml first thing — corgi create or /corgi-new — so "how do I run this?" has a permanent answer.
  • /corgi:stories — let Claude plan a feature across your services, back to front, and open a draft PR for each.

While in services you can create whatever you want, but in db services for now it supports:

Preflight & healthcheck​

  • corgi doctor (alias: check) — before corgi run: verifies every tool in required:, Docker is up, and every port in db_services.*.port / services.*.port is free (lists the offending process if not)
  • corgi status (aliases: health, healthcheck) — after corgi run: TCP-probes every declared port. If a service sets healthCheck: /some/path, corgi does an HTTP probe and accepts any non-5xx response as healthy. The localstack driver defaults to GET /_localstack/health.

Install​

After install, corgi is available globally — run it from any folder.

macOS / Linux — Homebrew​

brew install andriiklymiuk/homebrew-tools/corgi

macOS / Linux — install script​

No Homebrew? One-liner that picks the right OS/arch binary from GitHub releases:

curl -fsSL https://raw.githubusercontent.com/Andriiklymiuk/corgi/main/install.sh | sh

Installs to /usr/local/bin if writable, otherwise ~/.local/bin (auto-added to PATH for zsh/bash/fish).

Useful overrides:

  • CORGI_VERSION=1.10.0 — pin a version
  • CORGI_INSTALL_DIR=$HOME/bin — force a directory
  • CORGI_NO_MODIFY_PATH=1 — don't touch shell rc files

Windows — PowerShell​

irm https://raw.githubusercontent.com/Andriiklymiuk/corgi/main/install.ps1 | iex

Installs to %LOCALAPPDATA%\corgi\bin and adds it to your user PATH.

Verify​

corgi -h

corgi update (alias corgi upgrade) detects how you installed and uses the matching method to upgrade.

Try it with expo + hono server example:

corgi run -t https://github.com/Andriiklymiuk/corgi_examples/blob/main/honoExpoTodo/hono-bun-expo.corgi-compose.yml

Vscode extension​

We also recommend installing corgi vscode extension which has syntax helpers, autocompletion and commonly used commands. You can check and run corgi showcase examples from extension too.

Claude Code users​

This repo ships a Claude Code plugin so an AI agent can author your corgi-compose.yml, run corgi, and debug failures accurately.

/plugin marketplace add Andriiklymiuk/corgi
/plugin install corgi@corgi

Then in any project that has a corgi-compose.yml, Claude will recognize it and use corgi run / corgi doctor / corgi status instead of inventing its own commands. The skills cover the whole loop: /corgi-run boots the stack (or a slice) detached and waits until healthy; /corgi-debug diagnoses a stack local-first and pulls deployed logs for a bug on demand; /corgi-suggest proposes ranked, measurable product + engineering improvements; /corgi:stories turns a batch of tracker issues (or a feature description) into a draft PR per service; /corgi:review reviews those PRs/MRs against your repo's standards and posts inline suggestions. Plus /corgi-new to scaffold a fresh corgi-compose.yml and /corgi-describe for a service map. See AI agents for the full picture.

Services creation​

Corgi has several concepts to understand:

  • db_services - database configs to use when doing creation/seeding/etc
  • services - project folders to use for corgi. Can be server, app, anything you can imagine
  • required - programs needed for running your project successfully (node,yarn,go,whatever you want). They are checked on init

These items are added to corgi-compose.yml file to create services, db services and check for required software.

Examples of corgi-compose.yml files are in examples repo. You can also check what should be in corgi-compose.yml by running corgi docs. It will print out all possible items in corgi .yml file or you can go to corgi compose items doc to see what the syntax and possible values of corgi-compose.yml

After creating corgi-compose.yml file, you can run to create db folders, clone git repos, etc.

corgi init

If you want to just run services and already created db_services:

corgi run

Tip: there can be as many services as you wish. But create it with different ports to be able to run in all at the same time, if you want.

You can read of what exactly happens on run or on init to better understand corgi logic.

Good to know​

Prerequisites. You need git, and Docker (only if you declare db_services). Everything else is whatever your project lists under required: — corgi doctor checks those, and corgi doctor --fix can install them by running the install: commands you put in the file.

Private repos just work. corgi clones with plain git, so your existing SSH keys / credential helper are used as-is. No corgi-specific auth to set up.

Secrets stay local. corgi writes each service's .env (db credentials, ports, sibling URLs) and gitignores .env* + corgi_services/* on init, so secrets never get committed. Put your own keys in a service's env or a tier file like env/staging/web.env — also gitignored.

When a run fails. Port busy → corgi doctor names the holder, corgi run --kill-port frees it. Missing tool or Docker down → corgi doctor --fix. Clone failed → check your git access to that repo.

Scope. corgi is a local inner-loop tool — for running your stack on your own machine, not staging/production or deploys. A corgi-compose.yml runs its start/beforeStart commands on your machine, so only run files you trust (especially corgi run -t <url>, which downloads and runs a remote one). corgi mcp --http is unauthenticated unless you expose it via --tunnel (which adds a bearer token), and corgi tunnel URLs are public — shut them down when done.

Low lock-in. Your services stay ordinary git repos, your databases are standard Docker (corgi writes a plain docker-compose.yml per database under corgi_services/db_services/), and the wiring is just .env files. Stop using corgi and you keep all of it.

Next steps​

  • Run a branch or worktree — run any service on a git branch or external checkout for a single run, no corgi-compose.yml edit.
  • AI agents — drive corgi non-interactively from scripts or an LLM: JSON output, stable exit/error codes, detached lifecycle, and the Claude Code plugin.
  • corgi-compose.yml items — every field you can set.
  • Commands — full CLI reference.