1. Use a local AI agent

Use a local AI agent that can run terminal commands and open browser pages. The published launcher prepares an exact microfeed release in a private cache, then directs the agent to the same guarded management engine available in the microfeed source repository. You do not need to Git-clone or open that source repository yourself.

1. Open a local AI agent

Open Codex, Claude Code, Cursor, or another capable local AI agent that can run terminal commands on your computer and complete a browser handoff. The current folder does not need to contain source code.

Your computer needs Node.js 22.12 or newer with npm. The published launcher includes its own Yarn runtime and deployment source, so ordinary deployment does not require Git or Corepack.

2. Give the agent one command

Paste this prompt:

Deploy microfeed to Cloudflare.
Start by running `npx @microfeed/cli manage`, then follow its instructions until
deployment is verified.

The first run verifies and copies the exact release bundled with @microfeed/cli, then installs the locked dependencies in your operating system cache with the included Yarn runtime. This may use about 1.3 GB and take several minutes, depending on your connection and computer. Later commands reuse that workspace, while saved deployment state lives separately so a cache refresh cannot erase your site connections.

The command prints the exact deployment-skill and reference paths for the agent to read. The agent should discover your Cloudflare login, ask you to choose an account or existing compatible site when necessary, initialize only when you want a new site, and verify the finished deployment.

3. Complete browser handoffs

Stay nearby for these steps:

• Cloudflare opens a browser page so you can sign in and authorize Wrangler.
• The agent asks before reusing resources or making important choices.
• The finished deployment opens a private, one-time page where you create the dashboard password.

Never paste a Cloudflare token, dashboard password, or private password-setup link into the agent conversation.

Verify the result

Ask the agent to run npx @microfeed/cli manage status. The final report should show the site address, dashboard protection, D1 database, and R2 state. Open the public address and confirm the site loads. You can sign in to the dashboard and publish a test item yourself, or follow the post-deployment checklist to enable @microfeed/cli for the agent.

If setup stops partway through, tell the agent to continue the same microfeed deployment. Initialization is designed to resume without recreating resources.

Next: complete the post-deployment checklist.