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

# Quickstart

> Go from idea to shipped software in minutes with the Coding Agent.

This guide walks you through creating your first project on pre.dev and shipping it with autonomous agents.

## 1. Create your account

<Card title="Sign Up" icon="user-plus" href="https://pre.dev">
  Create a free account at pre.dev to get started.
</Card>

## 2. Start a new project

From the homepage, you have three ways to start:

| Mode                    | Description                                                     |
| ----------------------- | --------------------------------------------------------------- |
| **Full-Stack Template** | Start from a curated template with a pre-configured stack       |
| **Adaptive**            | Describe your idea and let pre.dev choose the best architecture |
| **Import**              | Connect an existing GitHub repo and continue building on it     |

Type your project idea into the input box. Be as descriptive as you want — the more context you give, the better the spec.

**Example:**

```
I would like to build a virtual museum tour platform with AR exhibits,
user-generated galleries, and a ticketing system for live events.
```

## 3. Choose your spec level

Before generating, choose between:

* **Plan** — A fast, high-level spec with milestones and user stories (\~30–60 seconds)
* **Deep Spec** — A comprehensive breakdown with granular subtasks and acceptance criteria (\~2–5 minutes)

See [Fast vs Deep Specs](/coding-agent/specifications/fast-vs-deep) for guidance on which to choose.

## 4. Review your specification

pre.dev generates a structured specification containing:

* **Technical architecture** and recommended stack
* **Milestones** — phased delivery plan
* **User stories** — feature requirements with acceptance criteria
* **Subtasks** — granular implementation steps (Deep Spec only)
* **Complexity estimates** per task

You can edit the spec, add/remove stories, or adjust the architecture before building.

## 5. Start building

Once you're happy with the spec, you have three build modes:

| Mode                   | How It Works                                                    |
| ---------------------- | --------------------------------------------------------------- |
| **Build Next Task**    | Agents build one task at a time, you review each PR             |
| **Build on Autopilot** | Agents work through the entire roadmap autonomously             |
| **Chat**               | Interactive mode — ask the agent questions or guide it manually |

Click **Build on Autopilot** to let agents ship the entire project, or **Build Next Task** for more control.

## 6. Review pull requests

Each completed task produces:

* A **feature branch** with the implementation
* A **pull request** for you to review
* **Passing acceptance criteria** (type checks, linting, browser verification)

<Note>
  The AI never pushes to main. You review and merge on your terms.
</Note>

## 7. Ship

Once you're satisfied with the code, merge the PRs and deploy. Your project is live.

***

## What's next?

<CardGroup cols={2}>
  <Card title="Build Modes" icon="play" href="/coding-agent/building/build-modes">
    Learn when to use Autopilot vs. manual build modes.
  </Card>

  <Card title="Integrations" icon="plug" href="/coding-agent/integrations/overview">
    Connect API keys, OAuth services, MCP servers, and skills.
  </Card>

  <Card title="Architect Agent" icon="compass" href="/architect-agent/overview">
    Use pre.dev's planner as an API inside your own coding agent.
  </Card>

  <Card title="Browser Agents" icon="browser" href="/browser-agents/overview">
    Scrape, automate, and extract from any web page.
  </Card>
</CardGroup>