Omnigent: The Open-Source Meta-Harness for Orchestrating AI Agents
Omnigent is an open-source meta-harness that unifies Claude Code, Codex, Cursor, and custom agents under a single orchestration layer with policy enforcement, sandboxing, and real-time collaboration.
What is Omnigent?
Omnigent is an open-source meta-harness for AI agents. Think of it as a universal remote for your coding agents: instead of juggling separate CLIs, configs, and credentials for Claude Code, Codex, Cursor, OpenCode, Hermes, Pi, and any custom agents you build, Omnigent gives you a single orchestration layer that ties them all together.
The core idea is simple but powerful: swap or combine agent harnesses without rewriting anything. You define your agents in YAML, set policies that govern their behavior, and run them from any device — terminal, browser, phone, or the native macOS desktop app. Sessions follow you, sub-agents collaborate, and everything stays in sync.
Why This Matters
If you've worked with multiple AI coding agents, you've felt the pain. Each one has its own CLI, its own credential management, its own sandboxing model (or lack thereof). Switching between them means context-switching. Combining them means duct-taping together scripts that barely work.
Omnigent solves this by providing a common orchestration layer that:
- Unifies multiple agents in a single session. Mix Claude Code, Codex, Cursor, and custom agents together. Ask one agent to review another's work, or split a task across agents that excel at different things.
- Works from any device. Start a session in your terminal, continue in the browser, pick it up on your phone. Messages, sub-agents, terminals, and files stay in sync.
- Enforces policies and sandboxing. Create policies that pause for approval before risky actions, cap spend, or limit which tools an agent can reach. Policies apply server-wide, per-agent, or per-session.
- Supports any model. Use a first-party API key, a Claude/ChatGPT subscription, or any compatible gateway (OpenRouter, LiteLLM, Ollama, vLLM, Azure).
Quick Start
Installation
One command installs Omnigent and everything it needs:
curl -fsSL https://raw.githubusercontent.com/omnigent-ai/omnigent/main/scripts/install_oss.sh | sh
Or install manually with uv:
uv tool install omnigent
# or: pip install "omnigent"
Homebrew users:
brew install omnigent-ai/tap/omnigent
Start Your First Agent
omnigent
This picks a model with you and starts a session in your terminal. It also launches a local web UI at http://localhost:6767 that shows the same session in the browser, or on a phone on your network.
You can also launch a specific agent runtime:
omnigent claude # Claude Code
omnigent codex # Codex
omnigent cursor # Cursor
omnigent opencode # OpenCode
omnigent hermes # Hermes Agent (Nous Research)
omnigent pi # Pi
Example Agents
Two example agents ship with the repo:
omnigent run examples/polly/
omnigent run examples/debby/
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.
Debby is a brainstorming partner with two heads — one Claude and one GPT. Every question goes to both heads, and she lays the two answers out side by side. Type /debate and the heads critique each other before converging.
Policy Governance
Policies decide what an agent may do: run shell commands, edit files, spend tokens. They check every action and either allow it, block it, or pause to ask you first.
In the web UI, open a session's info panel to browse available policies and toggle them on or off. In chat, just ask: "Add a policy that asks me before running shell commands." The agent sets it up for you.
For server-wide defaults, define them in your server config or an agent's YAML:
policies:
approve_shell:
type: function
handler: omnigent.policies.builtins.safety.ask_on_os_tools
cap_calls:
type: function
handler: omnigent.policies.builtins.safety.max_tool_calls_per_session
factory_params:
limit: 50
budget:
type: function
handler: omnigent.policies.builtins.cost.cost_budget
factory_params:
max_cost_usd: 5.00
ask_thresholds_usd: [3.00]
Policies stack across three levels — server-wide (admin), per-agent (developer), and per-session (you) — with stricter session rules checked first.
Writing Your Own Agent
An agent is a short YAML file: your prompt, your tools — local Python functions, MCP servers, and sub-agents a supervisor can delegate to.
name: my_agent
prompt: You are a helpful data analyst.
executor:
harness: claude-sdk
tools:
word_count:
type: function
callable: mypackage.mymodule.word_count
docs:
type: mcp
url: https://example.com/mcp
researcher:
type: agent
prompt: Search for relevant information and summarize it.
tools:
word_count: inherit
Run it with:
omnigent run path/to/my_agent.yaml
Deployment
Run Omnigent on a server with a stable URL and your sessions become reachable from anywhere, including your phone. One docker compose up runs the server on any host you have (VPS, home server). Render and Railway deploy with one click. Fly.io, Hugging Face Spaces, Modal, Cloudflare (serverless, scale-to-zero), and Databricks Apps are covered too.
Collaboration
Omnigent supports multi-user accounts. Enable authentication with:
OMNIGENT_AUTH_ENABLED=1 omnigent server start
Share a live session via the web UI, co-drive with teammates using omnigent attach <session_id>, or fork a conversation to continue independently with omnigent run --fork <session_id>.
Conclusion
Omnigent is still in alpha, but it's already a powerful tool for anyone who works with multiple AI coding agents. By providing a unified orchestration layer with policy governance, sandboxing, and real-time collaboration, it solves real pain points that developers face daily. The open-source nature means you can extend it, contribute to it, and shape its future.
Check out the GitHub repo to get started.