Agent setup.
Paste one prompt into Claude Code or Codex on your Mac; the agent installs the CLI, runs prized setup, and asks before putting your project or your credentials on the box.
One prompt
Onboarding builds this prompt for you: after your first box boots, the connect step asks setup's questions in the browser (your agent, then the project and how it reaches the box) and composes the prompt with those answers already carried as flags, so the agent has nothing to re-ask. The composed prompt includes the GitHub handoff (--gh) by default; the prompt spells it out, and you can delete that flag before pasting if you do not want gh on the box. Copy it there and you are done.
The version below is the generic template for every other time: a later machine, a fresh start, or when you would rather the agent relay your choices itself. Open your agent on the Mac, in the project you want on the box, and paste this. It covers steps 2 through 4 of the Quickstart; you still create the box in the dashboard if you have not already.
Set up Prized on this Mac and, if I confirm each part, put this project
and my coding agents on my box. Prized is a cloud dev box that syncs
folders with this machine; the docs live at https://prized.dev/docs
(plain-text mirror at /llms-full.txt).
Rules: copy, never move or delete; never print credentials or read
~/.config/prized; ask me before installing software, before transferring
any credential, and before doing anything that could cost money.
Non-interactive prized commands accept --json and print exactly one JSON
object with an "ok" field; error codes and exit codes are stable and
listed at https://prized.dev/docs/cli.
1. Install: if `prized version` fails, ask me, then run
`curl -fsSL https://prized.dev/install.sh | sh` and re-check.
2. Sign in: if prized commands fail with code AUTH or a "prized login"
hint, stop and ask me to run `prized login` in another terminal (it
needs my browser). I can instead paste a CLI token from the dashboard
for you to use as `prized login --token <token>`.
3. Everything else is one command, safe to re-run:
`prized setup --json --non-interactive`. It never prompts; each
consent-gated step runs only when its grant flag is present and is
reported as skipped otherwise. Ask me before adding each flag, and
only pass the ones I say yes to:
- `--project <path>`: sync this folder with the box in place; the
folder itself becomes the live side and the box wins conflicts.
Ask me first. Two alternatives are their own distinct grants, so
relay my choice exactly: `--copy` syncs a copy under ~/Prized and
leaves the folder alone; `--clone` clones the repo onto the box
over git with no live mirror, the right choice for a repo I
actively work in elsewhere. Say which one I picked in your
report; if I say no to all of them, run setup without --project.
- `--agent claude|codex|cursor|chatgpt` (repeatable): hand that
agent's sign-in and MCP config to the box. Ask me per agent.
- `--install-cursor`: run Cursor's installer on the box. Its own yes.
- `--gh`: copy my GitHub CLI token and git identity. Its own yes.
If setup fails with exit 5 naming multiple candidate boxes, ask me
which one and re-run with `--box <name>`; if it finds no box on the
account, ask me to create one in the dashboard. Read steps[] in the
JSON: "skipped" entries name what was declined or not granted,
"failed" entries carry the hint.
4. Claude Code sign-in is my step, not yours: it needs a browser
(`claude setup-token`), so a non-interactive run reports claude auth
as manual. If I picked claude, ask me to run
`prized setup --agent claude` myself in another terminal; the token
goes from my Mac to the box directly and you never see it. Then
continue.
5. Verify: `prized status --json` reports ok; if we set up the project
as a sync (not `--clone`; a clone has no sync session), also that
`prized sync flush <name>` completes. Check
`prized agents ls --json` shows each chosen agent installed and
signed in on the box.
6. Report: the box name, the project path on both sides (or the clone
path on the box, or that I chose to keep this repo local), which
project mode I granted, each agent's handoff outcome, and the
reattach command for any session setup started.What it asks you
Sign-in: prized login opens your browser, so the agent stops and asks. Run it yourself in another terminal, or mint a CLI token in the dashboard and hand it over for prized login --token.
The transfer: the agent confirms before it passes --project, so pasting the prompt in the wrong window never ships a project you did not mean to sync. The grant means sync in place: the named folder becomes the live Mac side of the pair, no copy, and the box wins conflicts in it from then on. The two alternatives are distinct grants the agent relays only when you pick them: --copy syncs a copy under ~/Prized and leaves your folder alone, and --clone puts a git clone of the repo on the box with no live mirror, which is the safe choice for a repo you actively work in elsewhere. Say no to all three and it still leaves you a signed-in CLI and a healthy box; add projects later with prized setup <path> or prized sync add. The modes are compared in Sync.
The credentials: in --non-interactive mode, prized setup skips every consent-gated step whose grant is missing, so the agent has to relay your yes as a flag: --agent <name> per agent, --install-cursor for the on-box installer, --gh for your GitHub token. One grant is not a flag: if you have picked agents on this Mac before, that choice is saved in [agents].preferred and a later run hands those agents off without --agent. That is the naming you already did, and it is why a re-run converges instead of asking again; pass --no-agents for a run that must depend on nothing but its own flags (the prompt the dashboard composes does exactly that when you pick "my agent is not listed"). Nothing else is copied that you did not name, the transfer runs over the SSH tunnel straight to the box, and the agent never handles a credential value; see Credentials.
Claude Code is the one sign-in the agent cannot do for you: claude setup-token needs your browser, so you run prized setup --agent claude in your own terminal when it asks.
Everything else is safe to leave to the agent. Every prized command in the prompt is safe to re-run, and a skipped or failed step is named in the JSON report rather than half-applied. The folder-pair model it sets up is described in Sync; for running agents on the box afterwards, see Agents.
Something unclear or out of date?