Getting Started

matic is a filesystem-first, git-native runtime for running structured teams of AI agents. Everything in the system, including orgs, projects, agents, tasks, and artifacts, lives as files in a git repo, operated through a CLI and coordinated by a persistent daemon. This section walks you from zero to a working org with agents executing real work, covering each primitive you need to understand along the way.

The guides below follow a natural progression: install the CLI, bootstrap an org, connect a channel, spawn agents, define a project, and execute your first run. If you want to skip straight to a working setup, start with Quickstart.

Installation

Install the matic CLI, configure prerequisites such as Bun, git, and agent runtime credentials, and verify your environment is ready in Installation.

Quickstart

Run matic quickstart to bootstrap a fully operational org in one command with Quickstart — scaffolded repo, default agent runtime, daemon running, sandbox mode enabled, and terminal chat ready.

First Org

Create an org repo from scratch in First Org, write a Charter that defines your operational scope and admissibility criteria, and configure the daemon that keeps your org alive.

First Channel

Connect the surfaces where signals enter the system in First Channel — terminal chat, Slack, Telegram, email, or webhooks — and configure how messages get routed to agents.

First Agent

Choose an archetype in First Agent, spawn a persistent agent, understand the onboarding sequence, and learn how agents accumulate identity, probes, memory, and skills across engagements.

First Project

Use First Project to write a proposal, define goals and milestones, staff a team, attach repos, and see how matic validates project scope against your org's Charter before work begins.

First Run

Follow a signal through the full operational loop in First Run: ingestion, activation, task decomposition, work item execution, HITL checkpoints, and artifact delivery.