Arc

Arc is an AI workspace with two modes:

• a main agent you talk to directly in chat
• optional task agents that pick up queued background work

You can start simple with one chat session, then add multi-agent automation only when you need it.

Fastest Setup

git clone https://github.com/mohit17mor/Arc.git
cd Arc
python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -e ".[dev]"
playwright install chromium
arc init
arc chat

When you want the dashboard, background task processing, and channels, start:

arc gateway

The dashboard runs at http://localhost:18789.

Start Here

• Getting Started: fastest path from install to first chat
• What Arc Is: the mental model in plain language
• Main Agent vs Task Agents: understand the two agent types
• Capabilities Overview: browser, memory, code tools, skills, MCP, and more
• Multi-Agent Overview: task agents, queues, chains, and reviews
• Providers Overview: global provider setup and per-agent overrides

What Arc Can Do

• Chat with a local or cloud model
• Run a web dashboard for chat, task management, logs, and agent management
• Queue tasks for named background agents
• Chain multiple agents in sequence with optional reviewers
• Use built-in tools like browser automation, code intelligence, memory, scheduler, and voice input
• Extend the system with custom skills or external MCP servers

Choose Based On What You Want

I just want a good local AI assistant

Start with Getting Started, run arc init, then use arc chat.

I want multiple specialized agents working in the background

Go to Multi-Agent Overview.

I want to connect more tools

Read Skills and MCP.

I want to tune providers and models

Read Providers Overview and Per-Agent Provider Config.