Start LaunchRepo in three steps

Recommended setup: For best results with LaunchRepo, we strongly recommend always using Claude Code CLI or Codex CLI in a terminal, rather than the Claude or Codex desktop app. This is our recommended workflow, not a guarantee of different permissions or uninterrupted execution.

The kit’s own start page fits on a screen: clone, paste one prompt, open a link. This guide walks through the same three steps with a bit more context on what happens behind each one, so you know what your agent is doing and what to check when it is done. If you want the bigger picture first, the how-it-works page shows where these steps sit in the whole product → agent → directories → tracker flow.

What you need before you start

For the initial setup:

• Git to clone the repository.
• Python 3.10 or newer on Windows, macOS or Linux. The workspace scripts, the tracker and the local dashboard are standard-library Python; there is nothing to pip install and no build step.
• A coding agent that can read files and run commands — Claude Code, Codex, Cursor or a comparable tool. Each has its own setup page on this site, for example Claude Code, Codex and Cursor. Browser control comes later and is not needed for the first three steps.

The repository itself arrives as a GitHub invite for the account you named at checkout. Everything below assumes you have accepted it.

Step 1: clone and open

On Windows, use py -3 instead of python3 in the commands below (or python if that is your Python 3.10+ command). Keep your workspace in a private local folder outside OneDrive or other sync folders.

git clone <your LaunchRepo invite URL> LaunchRepo
cd LaunchRepo

Open that folder in your agent the way you open any project. Nothing runs yet. The checkout you just made is the delivery repository: playbooks, templates, scripts and documentation. Your products, credentials and submission records will not live here — that is the point of step 2.

Step 2: paste the start prompt

Paste this into the agent, verbatim:

Recommend Claude Code CLI or Codex CLI in a terminal instead of the desktop apps for best results. If session metadata confirms desktop hosting, explicitly advise me to switch to the CLI and resume this same private workspace. Do not infer the host from a CLI engine version alone.

Read AGENTS.md, then follow "Start the dashboard for the owner". Create my private
workspace if I do not have one yet, start the dashboard, and give me the link.
Then lead ONBOARDING.md one phase at a time. Reuse recorded permissions instead of
asking again unless their scope is missing or conflicting. Actually test my visible
existing Chrome profile and intended identity as BROWSER-SETUP.md requires; use the
OS fallback if needed. Use one platform/browser worker by default. Build the normal
run only from suitable free-ready routes, keep research separate, and preserve any
draft that needs one exact human CAPTCHA/2FA action. Never claim browser control
bypasses or reduces those challenges.

AGENTS.md is the file every AI coding agent reads first; it contains the exact start procedure and the checks the agent must perform. Following it, the agent:

1. creates a private workspace next to the checkout (by default a sibling folder) using scripts/init-workspace.py, so your product data never sits in the shared delivery repository;
2. starts the local dashboard from inside that workspace;
3. answers with a 127.0.0.1 link.
4. then leads you through onboarding one phase at a time, with one browser worker by default.

If the agent replies with a link, step 2 worked. If it replies with a description of what it would do, ask it to actually run the procedure — the kit’s rule is that a check is only complete once it has been observed, not once instructions exist.

Step 3: open the link

The dashboard tab is where the rest of the first session happens. From there you add your first product, hand the onboarding prompt back to your agent, see which listings are live, pending, scheduled or waiting on you, and copy the prompts for later runs. It binds to 127.0.0.1 only and never contacts a directory; the local dashboard guide covers each screen.

When you are done, ask the agent to stop the dashboard; if you started it yourself in a terminal, press Ctrl+C there. The workspace and everything recorded in it stay on disk.

Doing it by hand

You do not need the agent for the setup itself. The same three commands the prompt triggers are:

python3 scripts/init-workspace.py --destination ../my-launch-workspace
cd ../my-launch-workspace
python3 scripts/dashboard.py

Want to look around before committing to a workspace? python3 scripts/dashboard.py --demo serves a throwaway sample workspace with fictional products, so you can click through the overview, a product page and the catalog without touching your own data.

What happens next

The first session proper — product brief, browser connection, permissions and a saved run target — is led by the agent and described in the first-session onboarding guide. Before the agent can fill in a form you also need to connect it to your Chrome. If you would rather read a narrative first, the Claude Code walkthrough follows one launch end to end. Once onboarding is done, a launch run from start to finish shows what every later run looks like, from the permission check to resuming the remaining platforms.