two80

https://280apps.com/setup.md - Agents start here

Commands

Every error prints a fix line with the exact command to run next.

| Command | Purpose | Options | | --- | --- | --- | | two80 | Show this directory's app state and next steps | | | two80 push | Build, deploy, and print the live URL | --name <slug>, --framework next\|static, --new | | two80 init | Detect the framework and write .280/config.json | --name <slug>, --framework next\|static | | two80 delete | Destroy this project's app, URL, content, and data | --yes <name> | | two80 whoami | Show authentication state | | | two80 login | Authenticate this machine | | | two80 setup | Install agent hooks and the skill | | | two80 version | Show the CLI version | --version, -v | | two80 help | Show all available commands and flags | --help, -h |

two80 push

Push initializes new projects automatically. Framework and name options only apply during the first initialization.

| Flag | Description | | --- | --- | | --name <slug> | Use this app name instead of the package.json name | | --framework next\|static | Skip framework detection | | --new | Create a fresh app instead of linking an existing one |

two80 push
two80 push --name my-app --framework next
two80 push --new

two80 delete

Deletion is permanent and requires the app's name as confirmation. Running the command without confirmation prints the required name and deletes nothing.

two80 delete
two80 delete --yes my-app

Agent integration

Two paths, install either or both:

• Session hook (recommended): npx -y two80@latest setup registers a SessionStart hook for Claude Code, Codex, and OpenCode. Every session opens with this directory's app state already visible. Live state, small per-session token cost.
• Skill: the same setup run installs the on-demand 280-deploy skill. Loads only when the agent recognizes a deploy task. No per-session cost, works in any agent supporting the skill format.

Both are opt-in, idempotent, and merge into your agent config without overwriting it.

Output contract

• stdout is TOON, for agents. Progress goes to stderr.
• Errors are structured on stdout and carry a runnable fix.
• Exit codes: 0 ok (including no-ops), 1 failure (with a fix), 2 bad flags or args.