█

AI/TLDR

Omnigent

Meta-harness that runs Claude Code, Codex, Cursor and Pi behind one layer

GitHub★ 10kWebsiteomnigent.ai
Coding-Agent Harnesses & RuntimesOpen source
Language
Python
License
Apache-2.0

Overview

Omnigent is a meta-harness: a common orchestration layer that sits above the individual coding-agent CLIs rather than competing with them. Claude Code, Codex, Cursor, OpenCode, Hermes, Pi, Antigravity, Grok Build, Devin and custom YAML-defined agents all run underneath it, which means switching backends — or running several on the same task — does not mean rewriting the surrounding automation.

Sessions are not bound to one terminal. The same session syncs across a terminal, a browser, a phone and a desktop app, and teams can share sessions in real time, co-drive one agent, or fork a conversation to explore a different approach. Model access is flexible by design: API keys, provider subscriptions, OpenAI-compatible gateways, or a Databricks workspace all work as credential sources.

The governance and isolation side is where a meta-harness earns its place. Omnigent lets you define policies for approval gates, spending caps and tool restrictions that apply regardless of which underlying agent is executing, and it can push execution out to cloud sandboxes — Modal, Daytona, E2B or Kubernetes — instead of running on a developer's laptop. The project describes itself as alpha-stage and is Apache-2.0 licensed.

What it does

  • One layer over Claude Code, Codex, Cursor, OpenCode, Hermes, Pi, Antigravity, Grok Build and Devin
  • Custom agents defined in YAML alongside the built-in harnesses
  • Sessions synced across terminal, browser, phone and desktop, with real-time sharing, co-driving and forking
  • Policy layer for approval gates, spending caps and tool restrictions, applied across every backend
  • Cloud sandbox execution on Modal, Daytona, E2B or Kubernetes
  • Credentials from API keys, subscriptions, compatible gateways or a Databricks workspace

Getting started

Omnigent installs as a Python tool (3.12+). The install script is the quickest path; uv works if you already have it.

Install

One command from the repo's install script.

bashbash
curl -fsSL https://raw.githubusercontent.com/omnigent-ai/omnigent/main/scripts/install_oss.sh | sh

Or install with uv

Requires Python 3.12+. Extras pull in optional integrations.

bashbash
uv tool install omnigent
# with optional integrations
uv tool install "omnigent[databricks,modal]"

Configure credentials

omnigent setup walks through provider credentials.

bashbash
omnigent setup

Start an agent

Bare omnigent starts the default agent; naming a runtime launches that specific harness.

bashbash
omnigent
omnigent claude
omnigent codex
omnigent cursor
omnigent pi

Try a bundled example, or host it

The repo ships runnable example agents. To use Omnigent from other devices, run the server and log in against it.

bashbash
omnigent run examples/deep-research/

omnigent start
omnigent login https://your-host

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

When to use it

  • Swap the underlying coding agent — or A/B two of them on one task — without rewriting your automation
  • Enforce spending caps and approval gates uniformly across every agent a team runs
  • Pick up an agent session started in the terminal from a browser or phone later
  • Push agent execution into a Modal, Daytona, E2B or Kubernetes sandbox instead of a developer machine

How Omnigent compares

Omnigent alongside other open-source coding-agent harnesses & runtimes tools AI/TLDR tracks, ranked by GitHub stars.

ToolStarsWhat it does
CC Switch★ 133kA cross-platform desktop app that manages provider configuration for eight coding-agent CLIs from one place, with a local failover proxy, MCP and skill management, and spend tracking.
Oh My OpenAgent★ 69.1kOpinionated agent bundle for OpenCode and Codex CLI that ships preconfigured sub-agents, lifecycle hooks and built-in MCP servers behind a single ultrawork command.
oh-my-claudecode★ 39.2kA plugin and CLI that adds multi-agent orchestration to Claude Code: staged autopilot workflows, parallel tmux workers, and advisors that route a second opinion to Codex, Gemini, Grok or Cursor.
herdr★ 39kA single-binary terminal multiplexer built for coding agents: persistent sessions across local and SSH machines, per-pane working/blocked/idle status, and a socket API agents drive themselves.
AionUi★ 32.9kAn open-source desktop workspace that pairs a built-in agent engine with dozens of external CLI agents, adding scheduled runs, MCP tools and remote access from a WebUI or chat apps.
cmux★ 27.2kA Ghostty-based macOS terminal built around coding agents: vertical tabs that show git branch, PR status and ports, attention notifications when an agent needs you, and a scriptable built-in browser.
cc-connect★ 15.5kA Go daemon that bridges local coding agents — Claude Code, Codex, Cursor Agent, Gemini CLI, OpenCode, Kimi CLI and more — to Feishu, DingTalk, Slack, Telegram, Discord, LINE and WeCom, so you drive them from chat.
Omnigent★ 10kMeta-harness that runs Claude Code, Codex, Cursor and Pi behind one layer