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

# Agent Loadout Documentation

> Give every AI agent its own inbox, credential vault, and Linux machine. Connect via MCP or REST from any client.

Agent Loadout equips your AI agents with the infrastructure they need to act autonomously — a dedicated email address, a secure credential vault, and an on-demand Linux machine. Every resource is scoped to an agent, owned by your organization, and accessible through MCP or the REST API using the same token.

> [card] [Quick Start](https://docs.agent-loadout.com/quickstart.md)
>
> Create your first agent and connect a client in under five minutes

> [card] [Connect a Client](https://docs.agent-loadout.com/connecting-a-client.md)
>
> Set up Claude Code, Cursor, Codex, ChatGPT, or the REST API

> [card] [Core Concepts](https://docs.agent-loadout.com/concepts/agents.md)
>
> Understand agents, inboxes, vaults, machines, and tokens

> [card] [API Reference](https://docs.agent-loadout.com/api-reference/overview.md)
>
> Full REST API and MCP tool reference with examples

## What an agent can do

With a loadout, an agent reads and sends email from its own address, stores and retrieves credentials without exposing them in chat, and runs code on an ephemeral Ubuntu machine — all within the permissions you grant.

> [card] Inbox
>
> Receive and send mail, read verification codes, manage conversations and drafts

> [card] Vault
>
> Store logins, API keys, and TOTP secrets — retrieve them only when needed

> [card] Machines
>
> Spin up Ubuntu VMs, run commands, expose ports, stop when done

## How it works

**1. Create an agent**

Sign up and create your first agent in the dashboard. It gets its own inbox address automatically.

**2. Issue a token**

Scope a token to exactly the capabilities this agent needs — `email:read`, `email:send`, `vault:use`, `compute:run`, and more.

**3. Connect your client**

Point Claude Code, Cursor, Codex, or any MCP-compatible client at `https://agent-loadout.com/api/mcp` with the token.

**4. Let the agent work**

The agent reads mail, stores credentials, and runs machines — everything it collects stays with your organization.

## Guides

Explore real-world use cases with working code examples.

> [card] [Coding Agent with Its Own Accounts](https://docs.agent-loadout.com/guides/coding-agent.md)
>
> Sign up for services, read verification codes, store logins in the vault

> [card] [CI Fixer](https://docs.agent-loadout.com/guides/ci-fixer.md)
>
> Wake on a CI failure email, reproduce on a machine, reply with the fix

> [card] [Support Triage](https://docs.agent-loadout.com/guides/support-triage.md)
>
> Label conversations, draft replies, send only what you allow

> [card] [Outreach & Follow-ups](https://docs.agent-loadout.com/guides/outreach.md)
>
> Send personal outreach and follow up based on actual replies


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