omnigent-ai

omnigent: Orchestrate AI Coding Agents Across Harnesses

Omnigent is a Python meta-harness for coordinating Claude Code, Codex, Cursor, and other coding agents through one shared layer. Run and govern agents from a terminal, browser, phone, or desktop app, and switch harnesses without rewriting agent configurations.

Stars
11k
Forks
1.8k
Stars, last 30 days
+486
Last commit
today
Contributors
100+

Our take

Newless than six months old and changing fast

Health score

82/100

How it is scored
Activity
80
Community
92
Issues
81
Pull requests
76

+486 stars in the last 30 days

  • Closes issues faster than 81% of the projects we track
  • Merges more pull requests than 79% of the projects we track
  • More contributors than 78% of the projects we track

Omnigent is a fast-moving, early-stage option for teams that need a shared control plane for multiple coding agents, with an active release and contribution history but limited time in the field so far.

Good fit if

  • You want to mix or change coding-agent harnesses without rebuilding your orchestration around each one.
  • You need session sharing, remote access, or policy controls for agent work.
  • You can work with a Python 3.12+ toolchain and are comfortable evaluating a young project.

Look elsewhere if

  • You only need one coding agent and do not need cross-harness orchestration or collaboration features.
  • Your environment cannot accommodate the Node.js requirements for harness CLIs and the web UI.
  • You require a mature, long-established project with a longer operational track record.
All health signals
Last commit2026-10-25 (today)
Commits, last 90 days100+
Releases, last 12 months22 (latest v0.19.1, 2026-10-21)
Contributors100+ (top contributor: 18% of commits)
Issues closed, last 90 days9+ (typically closed in 0 days)
Pull requests merged, last 90 days65 (typically merged in 0 days)
Project age5 months

Checked on 2026-10-25 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

11,482

Forks

1,933

Change since Nov 21, 2026: +10 stars·-1 forks

Overview

Omnigent is a Python meta-harness that provides a common orchestration layer for Claude Code, Codex, Cursor, OpenCode, Hermes, Pi, and custom agents. It addresses the problem of managing agents through separate tools: teams can combine or switch harnesses, apply shared policies, and continue sessions across a terminal, browser, phone, or native desktop app.

It is aimed at developers and teams that want to supervise multiple coding agents, run them on local or cloud hosts, and collaborate on their sessions. Agents can be defined in YAML, with tools and sub-agents, while policies can constrain actions or require approval.

Key Features

  • Run supported coding-agent harnesses and custom YAML-defined agents through a shared interface.
  • Coordinate agents in one session, including delegating work and routing changes for review.
  • Continue sessions across the terminal, web UI, phone, and macOS desktop app.
  • Share sessions for live collaboration, co-driving, or branching a conversation into a separate session.
  • Apply policies at server, agent, or session level to approve actions, limit tools, or cap spend.
  • Use local machines or configure supported cloud sandbox providers as agent hosts.
  • Connect API keys, subscriptions, compatible gateways, and Databricks credentials.

Use Cases

  • A developer can delegate parallel implementation tasks to different coding agents and review their changes before merging.
  • A team can share a live agent session so colleagues can observe, ask questions, or take over work on the host machine.
  • An organization can set approval rules and spending limits for agents used across a server or by specific agents.
  • A developer working across devices can start a session in the terminal and continue it from a browser or phone.
  • A team can run agent sessions in disposable cloud sandboxes when local machines should not remain online.

What you need

Detected in the repository

  • Node.js (from package.json)
  • Python >=3.12 (from pyproject.toml)
  • A test suite and automated checks on GitHub Actions

License in plain words

Apache-2.0permissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the license, the NOTICE file and a note of your changes
  • Share your changes: no
  • Includes an explicit patent grant from the contributors:

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Install with the project bootstrap script:

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

Then start Omnigent with omnigent or launch a specific harness, for example omnigent claude. See the README for prerequisites, configuration, deployment, and platform-specific details.

Alternatives

  • cli-agent-orchestrator: It coordinates coding CLIs in isolated tmux sessions under a supervisor, rather than providing a shared layer across harnesses and interfaces.
  • bernstein: It focuses on deterministic workflow scheduling, policy enforcement, approvals, and audit records for agent tasks.
  • ruflo: It is a TypeScript harness focused on Claude Code and Codex, with agent teams, memory, and MCP integrations.
ProjectLanguageLicenseStarsStatus
omnigentPythonApache-2.011kNew
cli-agent-orchestratorPythonApache-2.01.4kActive
bernsteinPythonApache-2.01.2kActive
rufloTypeScriptMIT73.8kActive

Considerations

Omnigent is marked alpha and is less than six months old, so its behavior and interfaces may change as it develops. The repository is changing quickly, with frequent commits and releases; contributor activity is broad rather than dominated by one person. The supplied health data also shows many open issues, so teams should assess unresolved problems relevant to their workflows before adopting it broadly.

The core package requires Python 3.12 or newer. Node.js 22 or newer is needed for coding-harness CLIs and the web UI. Native terminal wrappers require tmux, and Linux sandboxing requires bubblewrap; Windows has degraded isolation and does not support those native wrappers. Optional providers and integrations may require extra packages and their own setup. Review the deployment guide and the policy guide for operational and security details.

Found this useful?

Share it with someone who would like omnigent.

Comparisons