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

# Your first task

> Create a lane, start an agent, watch the tool calls, and review the diff.

This is the shortest path from an open repository to a reviewed change. It assumes ADE is installed and at least one provider is connected — see [Set up ADE](https://ade-app.dev/docs/getting-started/install.md) and [Connect a provider](https://ade-app.dev/docs/getting-started/connect-provider.md).

![ADE's Work tab with lanes and the composer](https://mintcdn.com/ade-ac1c6011/m_EKJqJVzAE9PL2s/assets/readme/start.gif?s=cbe5a09688f9443b4d9974062f00869c)

The Work tab: lanes on the left, the active chat and composer on the right.

## 1. Create a lane

A **lane** is ADE's name for a git worktree: an isolated branch and working copy for one task. It keeps agent work out of your primary checkout, so you can run several tasks in parallel without colliding.

> [!NOTE]
>
> Your repo root is already the **primary lane** — it always exists and is edit-protected. You don't pick a "lane type"; you add a new worktree lane for the task, branched off your default branch.

**1. Open Lanes**

Choose **Lanes** from the sidebar.

**2. Click New lane**

Click the **+** above the lane list. Give it a short task name, like `add-health-endpoint` or `fix-login-copy`.

**3. Pick the base branch**

For your first lane, use your default branch — usually `main`.

**4. Create**

ADE creates a Git worktree under `.ade/worktrees/` and checks out a branch for the lane.

### Or let ADE create it

Describe a task in chat and ADE spins up a fresh worktree, auto-named from the task — no stashing, no context switch.

![A lane selected in the Lanes tab](https://mintcdn.com/ade-ac1c6011/Mghbi7-G0yTaUGxT/media/current/lanes.jpg?fit=max\&auto=format\&n=Mghbi7-G0yTaUGxT\&q=85\&s=e5669a6b1bfe74167b72181ef6f9e3af)

A lane with its branch, open PR, CI state, and the chat that produced it.

### Good lane names

| Good | Avoid |
| - | - |
| `add-health-endpoint` | `test` |
| `fix-auth-redirect` | `agent-work` |
| `linear-eng-142` | `new-branch` |

## 2. Start the chat

**1. Open a new chat**

In **Work**, start a new chat in your lane. You can also let ADE create the lane from the prompt.

**2. Pick a provider and model**

Use the default, or choose any connected provider from the model selector. Thirteen runtimes are supported; see [Agents and models](https://ade-app.dev/docs/concepts/agents-and-models.md).

**3. Send a narrow task**

Ask for one contained change. Skip broad prompts like "clean up the whole app" for your first run.

## 3. Watch, review, and commit

The transcript streams reads, edits, shell commands, and test output as the agent works. Every tool call is expandable, so you can inspect each edit before you trust it.

When the agent finishes:

* Open **Files** and read the diff.
* Run the relevant checks from **Run** or the terminal.
* Read the closeout summary.
* Commit, or open a pull request if the change should be reviewed.

## First prompt

```text theme={null}
Add a GET /health endpoint that returns { "status": "ok" } and add one test for it.
```

## If the agent gets stuck

> [accordion] Unstick a run
>
> * Press **Stop** and ask for a shorter plan.
> * Paste the exact test or build error back into the chat.
> * Switch to a stronger model for the session.
> * Keep the task inside the lane — don't move it to your primary checkout.

> [!TIP]
>
> Your first successful session should be one lane, one agent, one narrow change, and one review pass. Build from there.

> [card] [Agent chat](https://ade-app.dev/docs/chat/overview.md)
>
> How chats, providers, and tool calls work.

> [card] [Pull requests](https://ade-app.dev/docs/tools/pull-requests.md)
>
> Open a PR once the lane is ready.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.