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

# Introduction

> Mains for AI coding agents

Mains is a desktop application for executing autonomous AI coding agents in secure workspaces, link your code repositories, import tasks from your PM tools, and watch agents handle reading, modifying, and submitting code on your behalf.

<Columns cols={2}>
  <Column>
    <Card icon="download" href="/install" title="Download">
      Get latest version of Mains for your Mac.
    </Card>
  </Column>

  <Column>
    <Card icon="square-q" href="/quickstart" title="Quickstart">
      First-time setup walkthrough
    </Card>
  </Column>
</Columns>

## Core Concepts

### [**Agent Workspaces**](/workspaces)

Run **Claude Code**, **OpenAI Codex**, **GitHub Copilot**, or **Cursor** agents inside [workspaces](/workspaces) linked to git repos. Each workspace gets its own Git worktree, so multiple sessions can run in parallel without conflicts. Agents can read files, run commands, write code, and track diffs.

### [**Linked Tasks**](/linked-tasks)

Connect issues and tasks from **GitHub**, **GitLab**, **Linear**, **Jira**, **Asana**, **Trello**, or **Sentry** directly to a workspace. The agent reads the task description as context, and statuses are updated automatically as work progresses.

### [**Code Reviews**](/reviews)

Request structured code reviews after agent runs. Findings are categorized by severity with file locations and suggested fixes.

### [**MCP Servers**](/mcp-servers)

Extend agent capabilities with [Model Context Protocol](/mcp-servers) servers. Add tools for Slack, databases, APIs, and more from the terminal -- Claude and Copilot agents can use them.

### [**Skills & Slash Commands**](/skills)

Teach agents domain-specific behavior with [skills](/skills), or create reusable [slash commands](/claude/slash-commands) for common workflows like code review, testing, and PR creation.

### [**Dependency Guards**](/dependency-guards)

Automatically check packages against [Socket.dev](https://socket.dev) before install commands execute. Guards intercept `npm install`, `pip install`, `cargo add`, and more across all three agent runtimes, blocking packages flagged for supply chain risks.

### [**Runs & Observability**](/runs)

Track every [agent run](/runs) with full observability -- tool calls, token usage, costs, and workspace diffs. Resume sessions, fork runs, and review artifacts produced by the agent.

### [**Pulse**](/pulse)

[Pulse](/pulse) runs agents on a schedule you define -- pick a workspace, prompt, and cadence for periodic reports, audits, and other recurring work without manual trigger.

### [**File Explorer**](/file-explorer)

Browse your repository and [select files](/file-explorer) to include in the agent's context. Combine file selections with linked tasks to give the agent a targeted view of both the requirements and the relevant code.

### [**Browser panel**](/browser)

Use the [embedded browser](/browser) in a workspace to open live pages, capture elements, and drop screenshots plus selectors into the agent's prompt for web-aware runs.

### [**Stats Dashboard**](/stats)

Track agent runs, costs, tool usage, and success rates across all workspaces. Filter by provider to compare Claude, Codex, Copilot, and Cursor activity.