Skip to content

Getting Started

This page is the shortest path to a working Arc setup.

You do not need to create task agents to get started. The main agent works immediately after arc init.

1. Install Arc

Requirements:

  • Python 3.11+
  • One model provider you can reach: Ollama locally, or a cloud API
git clone https://github.com/mohit17mor/Arc.git
cd Arc
python -m venv .venv

.venv\Scripts\activate

source .venv/bin/activate

pip install -e ".[dev]"
playwright install chromium

2. Run The Setup Wizard

arc init

The wizard sets up your main agent identity and your default provider.

It walks through:

  1. your name and the agent name
  2. the main agent personality or custom system prompt
  3. the default LLM provider and model
  4. optional extras like Telegram, Tavily, and ngrok

Arc stores this in ~/.arc/config.toml and ~/.arc/identity.md.

3. Start Your First Chat

arc chat

This is the easiest way to understand Arc. At this point you already have:

  • the main agent
  • your default provider
  • built-in capabilities
  • any local skills and MCP config you have added

4. Run The Gateway When You Need More Than Chat

arc gateway

Use arc gateway when you want:

  • the web dashboard
  • task processing for named task agents
  • Telegram or other connected channels
  • real-time logs and background services

Open http://localhost:18789 in your browser.

5. Optional Next Steps

Add named task agents

If you want background workers for specialized jobs, go to Create Agents.

Queue tasks and chains

If you want work to continue without staying in chat, go to Queue Tasks and Multi-Step Chains and Reviews.

Add more tools

If you want to extend Arc, go to Skills and MCP.

First-Day Mental Model

  • arc init configures the main agent and the default LLM settings.
  • arc chat talks to the main agent.
  • arc gateway turns on the dashboard and background systems.
  • Task agents are optional. You only create them when you want queued multi-agent work.