---
title: "Quickstart"
description: "Sign up, connect a GitHub repository, and start your first session in a few minutes."
---

> Documentation Index
> Fetch the complete documentation index at: https://docs.trevize.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart

This guide takes you from zero to your first agent-authored pull request. You'll
create a workspace, connect a repository, and start a session.

> **What you'll need**
>
> A GitHub account with access to a repository you can connect, and a Claude Code
> API token (Trevize uses it to run the coding agent).

1. **Sign up**

   Create an account and sign in. You can use **email and password**, or
   **Continue with GitHub** / **Continue with Linear** to sign in with an existing
   account. If you were invited to a workspace, enter your invite code during
   sign-up.

2. **Create your workspace**

   On first sign-in, the getting-started flow walks you through a short setup:

1. **Create workspace** — give it a name; Trevize generates a URL slug (your
   workspace address) from it.
2. **Choose your style** — pick a light or dark appearance (changeable later).
3. **Invite your team** — optionally add teammates by email (you can skip this).

   A workspace is your team's home: it owns your connected repositories, members,
   and billing. When setup finishes you land on your workspace home.

3. **Connect GitHub**

   From your workspace home, open the **Connect GitHub** card and install the
   Trevize GitHub App, then authorize the repositories you want to use.

   This is what lets Trevize **create branches, push code, and open pull requests**
   on your behalf — so agent work shows up as reviewable PRs in your repo. See the
   [GitHub integration](/integrations) for details.

4. **Set up Claude Code**

   Open the **Set up Claude Code** card and add your Claude Code API token. Trevize
   uses it to run the coding agent that powers sessions, tasks, and automations.

   Optionally, **Connect Linear** too, so you can turn issues into work and track it
   across tools.

5. **Start your first session**

   Click **New session** to open the *New collaborative session* dialog and fill in:

- **Title** — an optional name, e.g. "Checkout redesign".
- **Repositories** — the repo(s) the agent should work in.
- **Base branch** — the branch to start from (defaults to your repo's default).
- **Model** — the agent model to use.

   Choose **Create session** and you'll drop into the session, where you can chat
   with the agent and steer it turn by turn.

   > **Prefer to fire-and-forget?**
   >
   > Instead of a live session, use the prompt card on your workspace home to start
   > a **task**: pick a blueprint (bug fix, feature, or quick change), describe what
   > you want to build, choose a repository and model, and let it run autonomously.

6. **Review and merge**

   The agent works in an isolated sandbox, checks out your repository, and
   implements your request on a branch. When it's ready you get a **pull request**
   (and often a **live preview**) to review. Leave feedback, request changes, and
   merge when you're happy — just like any other contributor.

## Where to next

- **Concepts** — Understand sessions, automations, briefs, and more in depth.
[Explore concepts](/concepts).
- **Guides** — Step-by-step walkthroughs for sessions, automations, and reviews.
[Browse guides](/guides).
- **Configuration** — Set up hooks and previews with `.trevize/settings.json`.
[Configure your repo](/configuration).

Source: https://docs.trevize.dev/get-started/quickstart/index.mdx