OpenHarness is a native application that serves as a unified command center for managing and orchestrating various AI coding agents across multiple machines. It enables developers to integrate different agents and machines seamlessly, supporting diverse disciplines from software development to CAD, robotics, and music, all within a secure, end-to-end encrypted environment.

openharness

Active·★ 1.3k·MIT·Updated 2026-10-22

★ Essential★ Trending

OpenHarness (called Harness in the app) is a native desktop workspace for running coding agents side by side. The README says it works with Claude Code, Codex, Cursor, OpenCode, Devin, Amp, Copilot and more, and lets you link a laptop, a home server and a GPU box into one window with a password, with no SSH keys, Tailscale or port forwarding. Each agent runs in its own tmux pane, the app does not wrap the agent's CLI but reads transcripts and uses the vendor's own hooks, and every harness can get its own git worktree so your working copy stays clean. Keys are remappable in a single JSONC file, remote terminals stream text peer to peer, and the README says traffic is end-to-end encrypted so the relay forwards bytes it cannot read.

Install the desktop app for macOS or Linux, then on each machine run the installer (`curl -fsSL https://harness.autonomous.ai/cli/install.sh | bash`, then `harness login`, `harness remote-password set` and `harness start`) and link it from Machines in the app; the same installer provides `hn`, a terminal front end built on tmux's keys. A Claude Code plugin lets agents start, wait for and read other harnesses. The README also pitches use beyond code through a Store of 49 domain-specific harnesses for tools such as Blender, CircuitJS, MuJoCo, a game generator and a music live-coder, and a companion hardware device. It lists performance numbers measured on an M2 Max, which are the project's own. Everything is MIT licensed.

#AI Agents#Development Environment#Multi-Machine Orchestration#End-to-End Encryption#Native Application#Domain-Specific Tools#LLM Integration

Features

01Unified AI Agent Management: Manage and orchestrate various coding agents (Claude, Codex, etc.) side-by-side in one interface.

02Multi-Machine Orchestration: Seamlessly link and control agents across your laptop, home server, and GPU box without SSH keys or port forwarding.

03End-to-End Encryption: Ensures code, keys, and keystrokes are sealed and transmitted securely via ChaCha20-Poly1305 encryption.

04Native and High Performance: A fast, lightweight native application (not Electron) with low resource consumption, allowing agents to continue working in the background.

05Domain-Specific Harnesses (DSH): Extensible framework to integrate tools from diverse crafts like CAD, circuits, robotics, and music with live viewers.

Why choose it

+One window for many agents (Claude Code, Codex, Cursor, OpenCode and more) across several machines

+Does not wrap agent CLIs: real tmux panes, transcripts read and vendor hooks used, so the agents behave as they normally do

+A git worktree per harness keeps your working copy clean

+Encrypted peer-to-peer terminal streaming, with no SSH keys or port forwarding needed to link machines

Trade-offs

!Desktop app is for macOS and Linux only, per the README

!Setup on each machine pipes a remote install script into bash and needs an account login; read the script first

!The performance numbers (M2 Max, 16 terminals) and the encryption claim are the project's own

!The non-code harnesses (Blender, circuits, music) and the hardware device are extras you may not need

Maintenance check

ActiveChecked: 2026-10-22

Last commit: 2026-10-22
Commits, last 90 days: 100+
Releases, last 90 days: 30
Latest release: v1.3.64_web (2026-10-22)
Open issues: 63
Recent issues with at least one comment: 0/1

Commits per week, last 12 weeks

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

#1410 Headless sign-in for a screenless device (lampOS V2 lamp hand-overs)3 comments · 2026-10-19
#284 Desktop app SIGSEGV in Flutter raster thread ~6s after launch (1.1.83, macOS 27)3 comments · 2026-09-24
#1449 Devin Cloud: harness wait exits 3 (needs the person) when Devin has only finished, so a question and a done task look the same1 comments · 2026-10-20
#706 Arabic / RTL (BiDi) text is not rendered correctly in Harness1 comments · 2026-10-04
#379 Declarative flows for the Orchestrator: run a task graph from a YAML file1 comments · 2026-09-27

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

Compatibility

macOS

Operating System

Verified via docs

Linux

Operating System

Verified via docs

tmux

Runtime

Verified via docs

Ollama

AI Runtime

Verified via docs

MLX-LM

AI Runtime

Verified via docs

Quick start

$ curl -fsSL https://harness.autonomous.ai/cli/install.sh | bash

$ harness login

$ harness remote-password set

$ harness start

Use cases

↳Design: Utilizing agents to generate and refine 3D models and designs using tools like Blender and OpenSCAD with live interactive viewers.

↳Circuit Engineering: Designing and simulating electronic circuits, allowing agents to modify components and analyze performance with overlaid traces.

↳Robotics: Simulating robot movements in environments like MuJoCo, pinning specific moments, and comparing different operational scenarios.

↳Game Development: Agents creating playable games using engines like Godogen, supporting features like rewinding, retrying, and providing feedback.

↳Music Composition: Composing and performing music as code with tools like Strudel, enabling real-time manipulation of tracks and exporting audio.

Alternatives

⚡Top VS:

VSAgenticX★597→VSadk-python★22.0k→VSagent-framework★14.3k→VSPocketFlow★11.3k→

AgenticX★ 597

AgenticX is a unified, production-ready multi-agent platform — Python SDK + CLI (agx) + Studio server + Machi desktop app. Features Meta-Agent orchestration, 15+ LLM providers, MCP Hub, hierarchical memory, avatar & group chat, skill ecosystem, safety sandbox, and IM gateway (Feishu/WeChat).

Features

Why choose it

Trade-offs

Maintenance check

Most discussed open issues

Compatibility

Quick start

Use cases

Alternatives

Comments

Features

Why choose it

Trade-offs

Maintenance check

Most discussed open issues

Compatibility

Quick start

Use cases

Alternatives

Comments