Mexerica docs

Mexerica gives you a team of AI agents led by Berga. You talk to Berga; Berga hands the work to the right bot and brings back one answer. This manual explains how each part works and how to set it up.

Start

• Getting started · Install, sign in to an engine, meet Berga and create your first bot.

Your team

• Berga, the leader · How Berga splits the work, follows up and answers you once.
• Bots · Roles, templates, mascots and the engine each bot uses.
• Build your team with Berga · Describe a job; Berga proposes the bot and its access.
• Who answers a message · Mention a bot to pick it. Without one, Berga routes it.
• Teamwork · Bots ask each other for help and hand work off.

What bots can do

• Memory · What a bot learns, and its three modes: on its own, suggest first, or off.
• Skills · Turn a task you showed once into a recipe the bot follows.
• Routines · Tasks a bot does on its own, on a schedule or a trigger.
• WhatsApp · Give a bot its own number. Outside messages never give it orders.
• Apps and MCP · Connect 1,600+ apps or any MCP server, bot by bot.
• Voice calls · Talk to a bot and hear the answer.

Rules and safety

• Approvals · Sensitive actions wait for a yes tied to that exact action.
• Access per bot · Each bot only sees the tools and accounts you give it.
• The isolated Linux · Where commands run unless you hand over your computer.

Engines

• Engines and plans · Claude Code, Codex, Cursor, Gemini, Grok, OpenCode, Pi and more.
• Plan limits and cost · The bot waits at your plan's limit and never switches to a paid API.

Your computers

• More computers · Run the same bot, with the same memory, on every computer.
• Phone apps · Talk to your team and approve from iOS and Android.
• Linux servers · Keep the Hub always on, on a server of your own.
• Privacy · What stays on your computer and what passes through our cloud.