Skip to main content
The home screen is where you set up a session before it exists. Nothing runs behind it and no machine is booted. Your first prompt creates the session and starts its machine, and What happens when you start a session follows it from there. To go back to work you already started, find it in the sidebar.
The Alpha home screen with the sidebar collapsed to a menu button at the top left. In the centre, the Alpha by Lyzr lockup sits above the one-line greeting Still up, Alex? Same. What are we building?, its second half in a blue-to-pink gradient, with a glowing horizon curving behind the composer. Below it is the empty composer with the placeholder Describe what to build — / for commands, @ for files…. Its bottom row shows the paperclip, the agent chip reading codex, the model chip reading gpt-6.1-sol · high, the plan chip, the setup chip reading blank machine, the repos chip, the microphone and a dimmed round send button. Under the composer is a row of seven chips: Apps, Agents, Decks & docs, Images & video, Data, Automations and Research.

The home screen: a greeting, one composer, and seven ways to start.

​
Open the home screen

Alpha opens on the home screen. To come back to it from a session, do one of these: Opening the home screen creates nothing, so you can leave it again without a stray empty session in your list. The composer has focus from the start.

​
Set up the next session

The chips in the composer’s bottom row set up the session your next prompt will start. The agent, model and plan chips sit on the left. On the right, before Send, are the machine and the repositories.
A close crop of the home composer's bottom row. On the left: the paperclip button, the agent chip with the Codex mark reading codex, the model chip with a chip icon reading gpt-6.1-sol · high, and the plan chip with a drafting compass reading plan. On the right: the setup chip with a box icon reading blank machine, the repos chip with the GitHub mark reading repos, the microphone button and a dimmed round send button with an up arrow.

What runs is on the left; where it runs is on the right.

The repositories are deliberately not remembered. That way a repo you picked for one task can’t end up in an unrelated session later. If you connect GitHub from the Repositories popover, Alpha takes you to github.com and back, and your typed prompt, staged repos and name are all still there. Connect GitHub and add repositories covers the popover in full. Choose the agent and model explains the models and efforts. There’s no permissions chip. Every new session runs with full access inside its own machine. The agent installs packages and runs commands without asking, and nothing it does reaches your other sessions. If you type /permissions, Alpha answers in a banner: “/permissions isn’t offered — every session runs with full access.”
If you don’t see the agent chip or the microphone, your Alpha isn’t set up to offer both agents or voice input. Without the agent chip, sessions run the one agent your Alpha provides.

​
Start from a suggestion

The seven chips under the composer are starting points: Apps, Agents, Decks & docs, Images & video, Data, Automations and Research. Click one and its list opens over the row, so the page doesn’t move. Each row is a short line such as “Take bookings for my small business” or “Get a morning digest of AI news”.
The home composer showing the first suggestion's full prompt in grey: Build a portfolio website to show my projects, with an about page and a contact form, then publish it live. Its bottom row shows the paperclip, the chips codex, gpt-6.1-sol · high, plan, blank machine and repos, the microphone and a dimmed round send button. Below it, in place of the suggestion chips, the Apps panel is open, with the Apps icon and title in its header and a close X on the right. Its rows read: Build a portfolio site and publish it live, highlighted with a chevron; Track habits with streaks and reminders; Take bookings for my small business; Manage my team's time-off requests; Run an online store with Stripe checkout; Make a neon snake game for the browser.

The Apps list opens where the chip row was, and the highlighted row's prompt previews in the composer.

A suggestion is a first draft. Edit it to match your project before you send it, for example by naming your sailing school and its class types.

​
Use commands before the session exists

Type / in the home composer to open the same command palette you get inside a session. Here it acts on the session about to start. Most commands change a chip. A few start the session at once, with the command as its first message.
The home composer with a single slash typed and the command palette open above it, over the greeting, whose last letters, …ding?, show at the right. Each row shows an icon, the command and a description: /model, choose what model and reasoning effort to use; /skills, list skills the agent can use ($name mentions them); /review, review my current changes and find issues; /rename, rename this session; /new, start a new session; /delete, permanently delete this session; /fork, fork this session — clones the VM, files and conversation; /init, create an AGENTS.md file with instructions for the agent; /compact, summarize conversation to prevent hitting the context limit. The /init row is highlighted. The footer reads ↑↓ navigate, ↵ select, esc dismiss and, at the right, commands. The composer's bottom row shows the paperclip, codex, gpt-6.1-sol · high, plan, blank machine, repos, the microphone and the white send button.

The palette on the home screen acts on the next session.

With a repository staged, /review against main or /init is a complete first message. You don’t need a prompt as well. In a Claude Code draft, Claude Code’s own commands and your skills also start the session at once. Commands that need a running session say so instead of doing anything:
Not sure what the next session will start with? Type /status before you send.
The Next session card lists Name, Agent, Model, Permissions, Mode, Setup, Repos, Files, GitHub, Integrations and Secrets. The Agent row appears only when your Alpha offers both agents. Its footer reads “The session your next prompt starts. Nothing is running yet; token usage appears once it is.” Each card closes with the × in its corner. Every command is listed in Slash commands.

​
Mention files and skills before the session exists

The workspace doesn’t exist yet, so @ and $ list what it will start with. Picking a file inserts @path, so the agent knows exactly which file you mean from its first turn. The composer page covers mentions inside a running session, including @session: for your other sessions. Skills explains what a skill adds.

​
What happens to my unsent prompt?

Alpha keeps your unsent text in this browser, so a reload or a closed tab loses nothing you typed.
Anything you send from the composer lands in the session’s transcript, keys included. If you paste something shaped like a key, the composer warns you: “Keys typed here land in the transcript — add it under Secrets instead, and the agent gets it as an environment variable.” Secrets shows how.

​
Next steps

What happens when you start a session

The boot status lines, how the session gets its name, and what to do if the first send fails.

Setups: ready-made machines

Save a machine with your repositories cloned and installed, then pick it from the setup chip.

Choose the agent and model

How Codex and Claude Code differ, what each effort costs, and when ultracode is worth it.

Prompt Alpha well

Turn a suggestion into a first prompt that names the outcome and how to check it.