Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/omnigent-ai/omnigent/llms.txt

Use this file to discover all available pages before exploring further.

This guide walks you through installing Omnigent, launching your first agent session in the terminal, and opening the live web UI — all from a single command. By the end you’ll have a running Claude Code or Codex session, a browser view at localhost:6767, and two example multi-agent workflows you can poke at immediately.
omni is a short alias for omnigent. Both names are installed on your PATH and resolve to the same entry point — use whichever you prefer throughout this guide and beyond.
1

Install Omnigent

The quickest path is the one-line install script. It installs uv if it isn’t present, installs the omnigent package, and offers to install tmux and Node.js if they’re missing:
curl -fsSL https://raw.githubusercontent.com/omnigent-ai/omnigent/main/scripts/install_oss.sh | sh
Prefer to manage the install yourself? Choose any of the alternatives below:
# uv (recommended for manual installs)
uv tool install omnigent

# pip
pip install "omnigent"

# Homebrew
brew install omnigent-ai/tap/omnigent
After installation, confirm the CLI is on your PATH:
omnigent --version
2

Run your first agent

Launch an interactive session. Omnigent picks up any model credentials already in your environment (an ANTHROPIC_API_KEY, OPENAI_API_KEY, or a claude/codex CLI you’re already signed into) and offers one as the default:
omnigent          # interactive picker — choose your harness and model
omnigent claude   # start directly with Claude Code
omnigent codex    # start directly with Codex
As soon as the session starts, Omnigent also launches a local web UI at http://localhost:6767. Open it in a browser to see the same session — messages, sub-agent trees, terminal output, and files — synced in real time. On your local network, any device (including a phone) can reach the session at your machine’s LAN address, for example http://192.168.x.x:6767.
The macOS desktop app wraps the same web UI in a native window and adds OS notifications and a dock badge. Download it at omnigent.ai.
3

Try the example agents: Polly and Debby

Two example agents ship with the repo and make great first sessions. Make sure you’ve cloned the repository, then run:
omnigent run examples/polly/   # 🐙 multi-agent coding orchestrator
omnigent run examples/debby/   # 🟠🔵 dual-head brainstorming partner
You can also override the top-level harness without changing the YAML:
omnigent run examples/polly/ --harness pi
omnigent run examples/debby/ --harness openai-agents
🐙 Polly is a multi-agent coding orchestrator who writes no code herself. She plans, delegates work to coding sub-agents (Claude Code, Codex, or Pi) in parallel git worktrees, then routes each diff to a reviewer from a different vendor than the one that wrote it. You merge.🟠🔵 Debby is a brainstorming partner with two heads — one Claude, one GPT. Every question goes to both heads and the answers are laid out side by side. Type /debate and the heads critique each other for a few rounds before converging. Debby requires both an Anthropic and an OpenAI credential; see step 4 to add them.
4

Set up model credentials

Run the interactive setup wizard to add credentials, set defaults, or remove a provider:
omnigent setup
Omnigent supports four kinds of credentials:
KindWhat it is
🔑API keyA first-party vendor key for Anthropic, OpenAI, and similar providers
🎟️SubscriptionA Claude Pro/Max or ChatGPT plan, via the official claude / codex CLIs
🌐GatewayAny OpenAI- or Anthropic-compatible base_url and key (OpenRouter, LiteLLM, Ollama, vLLM, Azure)
🧱DatabricksA Databricks workspace profile (requires the databricks extra)
Defaults are per-agent, so a Claude default and a Codex default coexist. You can also switch models mid-session with the /model command.

What’s next?

Write Your Own Agent

Define a custom agent in YAML with your own prompt, tools, and sub-agents.

Govern Agents with Policies

Add approval gates, spend caps, and tool-call limits to any session.

Collaborate with Your Team

Share live sessions, co-drive, fork conversations, and manage multi-user accounts.

Deploy a Server

Run Omnigent on a VPS, Render, Fly.io, or Modal so sessions are reachable from anywhere.

Build docs developers (and LLMs) love