Vega

Guide

Getting started

From zero to a running AI agent in under 5 minutes.

Prerequisites

1

Install the CLI

# Homebrew

brew install everydev1618/tap/vega

# Or download a binary

macOS · Linux · Windows →

# Verify

vega version

vega v0.4.0 (go1.24)

2

Set your API key

# Create config directory and env file

mkdir -p ~/.vega

echo 'ANTHROPIC_API_KEY=sk-ant-...' > ~/.vega/env

Or export directly: export ANTHROPIC_API_KEY=sk-ant-...

3

Launch the dashboard

vega serve

Server started on :3001

Dashboard: http://localhost:3001

Open http://localhost:3001 in your browser. Two agents are ready to go:

Iris

Your chief of staff. Routes messages, manages MCP connections, triages agent requests.

Hera

The agent builder. Creates agents, teams, and channels on the fly via chat.

4

Build something

Talk to Iris in the dashboard. She'll figure out what you need and have Hera build it.

You: I need someone who can research topics and write blog posts about them

Iris: I'll have Hera set that up for you.

Hera: Done, love. Meet Sofia -- she's your Content Writer. You can find her in the sidebar.

Or define agents in YAML for reproducible setups:

name: MyApp

agents:

sofia:

model: claude-sonnet-4-20250514

system: You write engaging blog posts. Keep them concise and practical.

tools: [read_file, write_file]

vega serve myapp.vega.yaml

5

Add integrations

Connect external services via MCP. Talk to Iris or add them to your YAML:

You: Connect Composio so my agents can use Slack and Gmail

Iris: I'll need your Composio API key.

You: It's cmp_xxx...

Iris: Connected. 47 tools now available. Your agents can use Slack, Gmail, and more.

Connections persist across restarts. No need to reconnect.

What's next

Examples

Multi-agent workflows, MCP demos, team setups

Documentation

Full architecture guide and API reference

MCP integrations

Connect GitHub, Slack, databases, and 850+ services

Contribute

File issues, submit PRs, join the conversation