# Quickstart

This quickstart is the fastest working path for a new user. It demonstrates all four core strengths in one journey:

1. One interface across Claude and Codex
2. GitHub-native operational memory
3. Built-in orchestration tools (`plan`, `run`, `review`, `iterate`, `status`)
4. Full-auto execution settings with resumable runs

{% hint style="info" %}
Prerequisite: complete [Installation](/getting-started/installation.md) first.
{% endhint %}

## Step 1: Initialize Locus in Your Repository

```bash
cd /path/to/your-repo
locus init
```

Expected outcome:

* `.locus/` is created with `config.json`, `LOCUS.md`, and `memory/` (structured learnings)
* GitHub labels are created or verified (`locus:*`, `p:*`, `type:*`, `agent:managed`)
* Your repository is ready for issue-to-PR execution

Deep dive: [`locus init` reference](/cli-reference/init.md)

## Step 2: Set AI Model (Unified Interface)

```bash
# Start with Claude
locus config set ai.model claude-sonnet-4-6

# Switch to Codex later without changing workflow commands
locus config set ai.model gpt-5.3-codex
```

Expected outcome:

* You can switch provider/model by changing `ai.model`
* The same Locus commands (`plan`, `run`, `review`, `iterate`) continue to work

Deep dive: [Unified Interface Across AI Clients](/core-strengths/unified-interface.md)

## Step 3: Plan GitHub-Native Work

```bash
locus plan "Add /health endpoint with tests and API docs" --sprint "Sprint 1"
locus sprint show "Sprint 1"
```

Expected outcome:

* Locus creates a sprint milestone and issue set on GitHub
* Issues are labeled for priority, type, and execution order
* Sprint scope is visible to your whole team in GitHub
* `locus run` will auto-detect this sprint — no manual activation needed

Deep dives:

* [GitHub as Backend](/core-strengths/github-backend.md)
* [Sprints and Issues](/workflow-deep-dives/sprints-and-issues.md)
* [`locus plan` reference](/cli-reference/plan.md)

## Step 4: Execute with Built-In Tools

```bash
# Execute sprint tasks
locus run

# Review generated PRs
locus review

# Apply review feedback
locus iterate --sprint

# Check operational status
locus status
```

Expected outcome:

* `locus run` executes tasks and opens PRs
* `locus review` posts review feedback
* `locus iterate` applies fixes back onto PR branches
* `locus status` shows current sprint/issue progress

Deep dives:

* [How Locus Works](/workflow-deep-dives/how-it-works.md)
* [Built-In Tools](/core-strengths/overview.md)
* [Execution Model (Technical)](/workflow-deep-dives/execution-model.md)

## Step 5: Enable Auto-Approval Settings and Resume

```bash
locus config set agent.autoPR true
locus config set agent.autoLabel true

# If a run is interrupted or fails mid-sprint:
locus run --resume
```

Expected outcome:

* PR creation and lifecycle labeling are automated
* Interrupted execution resumes from the next unfinished task
* Completed tasks are not re-executed

Deep dives:

* [Auto-Approval Mode](/core-strengths/auto-approval-mode.md)
* [Execution Model (Technical)](/workflow-deep-dives/execution-model.md)

## Complete Command Block

Use this when you want the full journey in one copy-paste sequence:

```bash
cd /path/to/your-repo
locus init
locus config set ai.model claude-sonnet-4-6
locus plan "Add /health endpoint with tests and API docs" --sprint "Sprint 1"
locus run
locus review
locus iterate --sprint
locus status
locus config set ai.model gpt-5.3-codex
locus config set agent.autoPR true
locus config set agent.autoLabel true
# Use only when a previous run was interrupted:
locus run --resume
```

## Where to Go Next

1. [How Locus Works](/workflow-deep-dives/how-it-works.md)
2. [Unified Interface Across AI Clients](/core-strengths/unified-interface.md)
3. [GitHub as Backend](/core-strengths/github-backend.md)
4. [GitHub-Native Workflows](/workflow-deep-dives/github-native-workflows.md)
5. [Built-In Tools](/core-strengths/overview.md)
6. [Auto-Approval Mode](/core-strengths/auto-approval-mode.md)
7. [Execution Model (Technical)](/workflow-deep-dives/execution-model.md)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.locusai.dev/getting-started/quickstart.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.