App Composer

Get ready.
Then make something.

A working environment comes first. Your agent handles setup, builds your first Hello app and checks that it works online. You handle sign-in and choices that need your say.

Step-by-step help

A recommended path and the next useful action. Optional details stay available.

Already have an agent report? Open or resume it here.

Where should your first app live?

Start with a small Hello app. No application idea required. Your agent reuses what you already have and prepares the missing pieces.

Choose a first address

Use the existing supported Cloudflare account. A free address does not make usage unlimited; your agent should review the actual account limits.

Your choice is kept on this device when saved. It grants no account or DNS access.

What do I need to do myself?
  1. Open a folder in Codex.

    Use an empty folder or your intended project. You do not need to set up a repository yourself.

  2. Give your agent the setup request below.

    The agent checks tools, connects your GitHub and Cloudflare accounts, prepares your private starter and verifies an HTTPS deployment.

  3. Complete official sign-in when asked.

    Keep codes and credentials in the provider’s login screen. Tell the agent when you’ve finished so it can continue.

Open the illustrated setup guide

Give your agent the next move.

Copy this request into Codex in your chosen folder. The agent should do the work, give you the smallest sign-in step when needed and return a redacted progress report.

A read-only check. Find missing tools or access without installing, creating repositories or deploying.

Read or copy the agent handoff
Check my environment without changing it. Use App Composer’s current Get Ready implementation at https://github.com/loftwah/app-composer. Read its trusted AGENTS.md and supported setup instructions in README.md; use the one lib/get-ready engine, not a new checklist or planner. Work in this authorised Codex folder. If a Get Ready MCP is already connected for this exact host-selected workspace and target, its check/inspect tools can perform read-only verification; they cannot execute setup. Otherwise use the same engine through the doctor command. Before changing an actual output project, use the same core’s inspectProject API or --inspect-project mode (or its host-selected get_ready_project tool). Inspect that project’s declarations, existing guidance and canonical skills without replacing them. Check user/plugin copies before installing duplicates. A file or matching entrypoint does not prove installed version, agent visibility, exercised capability or readiness; recheck the output project’s actual environment separately. Keep credentials out of chat and specifications, preserve existing files and save resumable evidence privately. Report what you actually checked, the next step and any precise human action. Give me a redacted progress JSON export using the doctor’s --report flag; never share its private receipt. Run the read-only doctor against this actual workspace, using only my explicitly selected account and existing deployment target if available. Do not install tools, create repositories, change origins or DNS, or deploy. Do not select another person’s account or domain. Separate missing tools, human sign-in, denied access, provider interruptions and unsupported cases. Continue independent checks and report their actual results; saving the private receipt is allowed. Explain the next safe step without executing it.
Which agents and machines work today?

The verified setup path uses Codex on an Apple Silicon Mac, with Node 22 and Git available. It can prepare missing GitHub CLI, pnpm and local Wrangler. First-account creation, global Git/Node installation and other runtimes still need additional support.

Another authorised agent can use the reusable library or the optional local MCP for checks. That does not certify its installation or grant setup permissions. Integration details.

Reference recipe codex-github-cloudflare-hello-v2. Reviewed tools: Wrangler 4.143.1, EmDash 1.2.0. Your agent checks the actual selected workspace; these are supported versions, not a claim about your computer.

Get Ready handles the environment before your idea. A saved address is a preference; your agent still checks account access, current source and HTTPS. You can pause after preparing the first app.

Check the project you want to build

Start with what’s already there.

Your project may already have instructions, design decisions and agent skills. Ask the agent to inspect that folder before adding or replacing anything.

Read on this device. Never uploaded or saved. This is a project-inspection report, separate from the environment report and private workspace receipt.

Prepare missing EmDash site skills

Your agent can add the reviewed site-building and CLI guidance for a supported EmDash project. Existing copies stay intact. It then needs to check that it can find and use the skills.

Copies a request for your selected project. It runs nothing here. Supported preparation and next checks

Library, CLI and MCP instructions

Check a domain or nameserver change

Your address, explained.

Your agent can check the exact domain in your Cloudflare account, compare assigned and observed nameservers, and inspect DNSSEC records. It changes nothing.

Read on this device; never uploaded or saved. Use the domain report, not a private workspace receipt.

Library, CLI and MCP instructions · Nameserver help

Keep what you build.

Your first app is a small, configurable Hello page with its own source. Keep it, change the greeting or use the verified environment for something bigger.

The launch reference belongs to this operator. Opening it does not verify your environment.

  1. Agent & tools

    A working folder and the tools to build.

  2. Your GitHub source

    A private starter you own and can update.

  3. Your Cloudflare account

    The intended account and deployment access.

  4. A live HTTPS app

    The exact source, deployed and verified.