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.

Omnigent ships as a standard Python package with two CLI entry points — omnigent and the short alias omni — and a set of optional extras for cloud sandboxes, tracing, and model providers. This page covers every supported install method, the prerequisites each one needs, and how to verify and upgrade your install.
Both omnigent and omni are installed on your PATH and resolve to the same entry point (omnigent.cli:main). There is no functional difference between them — they are interchangeable aliases. Do not install both packages separately; a single pip install omnigent (or equivalent) provides both names.

​
Prerequisites

Before installing, make sure the following are available on your system:
  • Python 3.12 or newer — Omnigent requires Python 3.12+. Check with python3 --version.
  • uv (required) — Omnigent’s install script and recommended install path use uv. Install it from docs.astral.sh/uv if it isn’t already present; the one-line install script offers to do this for you.
  • git (required) — used for cloning agent repositories and the examples.
  • Node.js 22 LTS or newer with npm (for Claude Code, Codex, and Pi harnesses) — omnigent run installs the harness CLI you pick on first use. See docs.npmjs.com.
  • tmux (for native wrappers) — required by the omnigent claude and omnigent codex terminal wrappers. Install with brew install tmux (macOS) or apt install tmux (Debian/Ubuntu). The one-line install script offers to install it for you.

​
Install Methods

The recommended path for most users. One command installs Omnigent and everything it needs, including uv and tmux if they are missing:
curl -fsSL https://raw.githubusercontent.com/omnigent-ai/omnigent/main/scripts/install_oss.sh | sh
The script is idempotent — running it again upgrades Omnigent to the latest release.

​
Optional Extras

Omnigent’s core install includes the Claude Code and OpenAI Agents harnesses. The following extras are opt-in and only needed for specific providers or features:
ExtraPackage(s)When to use
databricksdatabricks-sdk, databricks-mcp, mlflow, psycopg[binary], opentelemetry-distroUsing a Databricks workspace as your model provider or running managed Databricks deployments
modalmodal>=1.0,<2Launching agent sessions in disposable Modal cloud sandboxes
daytonadaytona>=0.180,<1Launching agent sessions in Daytona cloud sandboxes
tracingmlflow>=3,<4Enabling MLflow tracing for agent runs (disabled by default; the server degrades gracefully without it)
bedrockboto3>=1.30,<2, botocore>=1.30,<2Routing model calls through AWS Bedrock
vertexgoogle-auth>=2.0,<3Routing model calls through Google Vertex AI
alldatabricks-sdk>=0.56.0,<1CI and dev installs that need the Databricks SDK alongside the default packages; also used by deploy tooling and Databricks MCP servers
Install one or more extras by appending them in brackets: 
uv tool install "omnigent[modal,tracing]"
# or
pip install "omnigent[databricks,bedrock]"

​
Verify the Install

After installation, confirm both CLI names are available and show the same version: 
omnigent --version
omni --version
Both commands should print the same version string. If either is not found, make sure the directory printed by uv tool dir (typically ~/.local/bin) is on your PATH.

​
Upgrading

uv tool upgrade omnigent

Build docs developers (and LLMs) love

Get started for freeTalk to us