dartway_cli 0.26.1

DartWay command-line tool: print the agent setup brief, check prerequisites, create projects from the canonical template, install the AI toolkit, run convention checks, deploy.

example/README.md

Example — from nothing to a running fullstack app

dart pub global activate dartway_cli
dartway create my_app

You get three packages — my_app_shared, my_app_server, my_app_flutter — and the agent toolkit in .claude/.

cd my_app/my_app_server
docker compose up -d          # Postgres and RustFS
dart pub get
dart run bin/server.dart      # migrates itself, then serves — nothing else to run first

# in another terminal
cd ../my_app_flutter
flutter run

Register from the app with your own phone number; the one-time code is printed to the server console. Your administrator identifier is set through DW_ADMIN_IDENTIFIER in deploy/secrets.yaml (git-ignored) — the role is granted by an admin, so the first one is declared per environment, not shipped in the skeleton.

What you get is a skeleton, not somebody's product: sign-in by a one-time code, a profile with roles, navigation with zone guards, an admin panel, a UI kit as source you own — and no domain models, because the domain is the part you write. The full path, command by command: quick start.

The other commands

dartway check     # the conventions, enforced: layout, generated code, migrations
dartway test      # server acceptance tests against a throwaway Postgres and storage
dartway stats     # code size per feature — what actually grew this week
dartway setup-ai  # install or update the AI toolkit (.claude/) in an existing project

Options worth knowing

dartway create my_app --channel master        # the development trunk instead of `stable`
dartway create my_app --local-repo ../dartway # a local monorepo checkout, no clone
dartway create my_app --no-git                # skip the initial commit

dartway check is the same checker an AI agent runs before it claims to be done — see the CLI.