Harness by Autonomous
Every coding agent, every machine, one command center.
508 followers
Every coding agent, every machine, one command center.
508 followers
Harness is the open-source command center for coding agents. Claude Code at work, Codex at home, Hermes in the cloud: every agent on every machine, side by side. Use the native desktop app, hn in your terminal, or the iOS and Android app to answer agents by voice. End-to-end encrypted. Then go beyond code with 49 harnesses for CAD, circuits, robots, games and music. Free on Mac and Linux.
Free
Launch Team / Built With
AWS Activate CreditsCredits that fuel your startup’s next milestone
Promoted
Hey Product Hunters! 🖐️
I'm Dee, Head of AI & Robotics at Autonomous. We're excited to launch Harness on Product Hunt today and show you what we built.
We run a lot of coding agents. Claude Code at work, Codex at home, Hermes in the cloud. One thing bothered us: every agent sits in its own window on its own machine, stops to ask something, and just waits.
We decided to change this. What if all of them lived in one command center?
🖥️ Every agent, every machine, side by side in a native app for Mac and Linux
🌿 Each agent on its own branch. Close the app and your agents keep working
🔒 End-to-end encrypted. Free and open source
🧠 Works with Claude Code, Codex, Cursor, OpenCode, Pi, Hermes and more
📱 Answer your agents on the go with the Harness app, available now on both the Apple App Store and Google Play Store. Talk instead of typing.
🎁 Exclusive to the Product Hunt community: 20% off Harness devices, the always-on display by your keyboard, with code PRODUCTHUNT.
We'd love to hear your honest feedback. What's your setup? It will help us improve and build more for everyone. 🙌 🙌 🙌
@dee_hw Aww man, this isn't fair. Came because of Harness but now I want to buy about every of the cool products you guys manufacture 🙄
I was confused at first but I extracted this "onboarding" from the main page without adding the harness and my agent almost bought a key (cute)
@christianruink Thanks, Christian, and sorry for the confusion. That's on us, not you. Love that your agent almost bought a key though 😄.
Quick version:
Harness runs the agents you already have, with your own accounts. So install the agent CLI first (Claude Code, Codex…).
⌘N, pick your agent and a folder, type the task, hit New Harness.
The pane opens with that agent running. If it isn't signed in yet, it shows the agent's own login right there.
Sign in once and you're set.
Where did it first feel confusing? We're fixing onboarding this week and your answer helps a lot.
congrats on the launch!!🚀🚀
the phone part is what i want. my org blocks the official remote thing for claudCode, so right now if an agent stops to ask something while i'm away it just sits there untill i get back.
can the relay be turned off entirely so it's machine to machine only?? thats probably what i'd need to show our security people.🤔
@seenew Thank you! 🚀 Remote approvals with your phone are exactly what it's for.
- Today, the relay can't be turned off. It handles sign-in, finding your machines, and connection setup, and it forwards traffic when a direct connection isn't possible.
- Content is end-to-end encrypted, always on, with no switch. Keys are generated on each of your devices and never leave them, using Ed25519, a PAKE for pairing, X25519, and ChaCha20-Poly1305. The relay sees metadata (machine names, agent names, online state, counters), never your prompts, code, transcripts or keystrokes.
- Terminal traffic goes direct. It moves to a direct WebRTC peer-to-peer channel once one can be established. Commands and events still pass through the relay, encrypted.
- For your security team: it's all MIT open source, including the relay (backend/), and the CLI, app and phone can point at a different backend URL. Self-hosting still depends on our sign-in today, so it isn't fully standalone yet. Start with docs/architecture.md and SECURITY.md in the repo: github.com/autonomous-ai/openharness. We're happy to answer their questions directly.
How does it handle machines that aren't always on? I move
between a desktop and a laptop, and every multi-machine setup
I've tried breaks the moment one of them is asleep.
@ankushchauhandev Chauhan, Good question. Harness has no hub machine. Each computer connects to our relay on its own, so a sleeping laptop never takes the desktop down.
- While one is asleep: it shows as Offline on the other machines and on your phone (detected within about 45s). Everything on the awake machine keeps working normally.
- When it wakes: it comes back online on its own, and the app reattaches to the agent you had open.
- The limit: agents run on the machine that started them, so an agent on a sleeping laptop pauses until the laptop wakes. Nothing migrates it to the desktop.
- For work that should never stop: start those agents on the always-on machine (harness new codex --machine desktop …). Then you can watch and steer them from the laptop or phone, no matter what state the other machine is in. Harness service install keeps the daemon running across reboots.
@dee_hw That's the right shape — no hub means a sleeping machine is
just offline, not a broken setup.
The "start them on the always-on machine, steer from the
laptop" pattern is the useful bit. That's how I'd end up using
it, and it's worth putting on the page — it isn't obvious
until someone spells it out.
This is interesting, I have organised all my agent windows together, works like a charm for complex workflows, though I have a doubt, whether it supports multiple accounts/profiles.
Suppose, If I'm working on a personal project using my personal claude/ codex account and at the same time I'm also working on my company's project using the organization's claude account? I have not tried it yet. So would be great if it does.
@ruffledzest Thanks! It depends on the engine:
- Codex: yes, per agent today. Each agent can use its own Codex profile folder (CODEX_HOME). The app finds folders like ~/.codex_work, or you can link one, so a personal and a company Codex agent can run side by side on one machine.
- Claude Code: not per agent yet. Agents use the Claude config the daemon starts with (your default login, or CLAUDE_CONFIG_DIR). Per-agent Claude accounts are a clear next step; your use case is exactly why.
- Workaround for now: run the company work on a separate machine or OS user signed in with the org's Claude account. Each one shows up in the same Harness.
Where do I enter the code at checkout??
@glacer06 So glad you're grabbing one, Nick! 🙌 At checkout, scroll to the bottom and, next to the terms line, enter PRODUCTHUNT in the promo code field for 20% off. Enjoy!
@dee_hw Done grabbed mine.
@glacer06 Thanks for the order, Nick! Excited for you to try Harness Device 🙌
this looks pretty interesting! I have a question pls, does it work good with agents running on remote devices?
@sig_eternal Yes, that's the main use case. Install Harness on the remote machine and sign in, and its agents show up next to your local ones. No SSH keys or VPN to set up. Harness connects your machines directly over WebRTC when it can, so typing in a remote pane feels local, and everything is end-to-end encrypted.
@sig_eternal One more detail: on the remote machine, run the 4 lines from our README (install, harness login, harness remote-password set, harness start), then go to Machines → Link Machine in the app.
github.com/autonomous-ai/openharness