Harness like a boss.

Claude Code at the office. Codex on the Mac mini at home. Hermes keeping your cloud server in shape. All your agents. All your machines. Now in one command center.

Harness
MACHINES
Claude Code·auth-apimacbook-pro
> refactor authentication to use the session cache
✻ Cooking… (4m 12s · esc to interrupt)
⏵⏵ auto mode on (shift+tab to cycle) · ● high · /effort
Codex·session-migrationoffice-desktop
add an index on sessions(user_id, expires_at)
Read schema.prisma
Wrote prisma/migrations/0043_add_idx.sql
Ran npx prisma validate
✓ Migration ready to review
gpt-5.6-sol default · ~/work/api
Pi·nightly-backuphome-server
> check tonight’s backup and summarize the result
· 18m
Hermes·server-healthcloud-server
check the server’s disk usage
/var/log is using 28 GB
Archived logs account for 21 GB
? Check which archives can be removed?
Waiting for your instruction

All your machines.
One command center.

The laptop in front of you. The server at home. The workstation at the office. The machine you rent in the cloud. Link each one with a remote password and open its agents here, side by side. No SSH. No open ports. No VPN.

Harness · everywhere
MACHINES
Claude Code·auth-apimacbook-pro
> refactor authentication to use the session cache
✻ Cooking… (4m 12s · esc to interrupt)
⏵⏵ auto mode on (shift+tab to cycle) · ● high · /effort
Cursor·sign-in-uimacbook-pro
> add the sign-in form and error states
Read src/SignIn.tsx
Edited src/SignIn.tsx +41 −9
? Use the API error text in the form?
Waiting for your instruction
→ Add a follow-up
GPT-5.6 Sol Extra High Fast · 5%
/ commands · @ files · ! shell
Pi·nightly-backuphome-server
> check tonight’s backup and summarize the result
· 18m
Hermes·ops-noteshome-server
summarize changes across the home services
Read the service notes
Ready for the next task

All your agents.
One command center.

Claude Code. Codex. Cursor. Hermes. Pi. Each in its own terminal, on the machine you choose. Writing code, running backups, keeping a server happy. Same agents. Same workflows. Or bring your own.

Harness · four engines, four panes
Claude Code·auth-apimacbook-pro
> refactor authentication to use the session cache
✻ Cooking… (4m 12s · esc to interrupt)
⏵⏵ auto mode on (shift+tab to cycle) · ● high · /effort
Codex·session-migrationoffice-desktop
add an index on sessions(user_id, expires_at)
Read schema.prisma
Wrote prisma/migrations/0043_add_idx.sql
Ran npx prisma validate
✓ Migration ready to review
gpt-5.6-sol default · ~/work/api
Pi·nightly-backuphome-server
> check tonight’s backup and summarize the result
· 18m
Hermes·server-healthcloud-server
check the server’s disk usage
/var/log is using 28 GB
Archived logs account for 21 GB
? Check which archives can be removed?
Waiting for your instruction

⌘B.
Boss mode.

Say what you want. Harness finds the agent already on it, wherever it’s running. No hunting for the machine. No hunting for the terminal. A clear match gets the work. A close call? You pick.

Harness
MACHINES
Claude Code·auth-apimacbook-pro
> refactor authentication to use the session cache
✻ Cooking… (4m 12s · esc to interrupt)
⏵⏵ auto mode on (shift+tab to cycle) · ● high · /effort
Codex·session-migrationoffice-desktop
add an index on sessions(user_id, expires_at)
Read schema.prisma
Wrote prisma/migrations/0043_add_idx.sql
Ran npx prisma validate
✓ Migration ready to review
gpt-5.6-sol default · ~/work/api
Pi·nightly-backuphome-server
> check tonight’s backup and summarize the result
· 18m
Hermes·server-healthcloud-server
check the server’s disk usage
/var/log is using 28 GB
Archived logs account for 21 GB
? Check which archives can be removed?
Waiting for your instruction

⌘S.
Swarm mode.

Multiple agents, one screen. Pick a layout and every session snaps into place. Home, office, cloud, side by side. See who’s working, who’s finished, and who needs you.

Claude Code·auth-apimacbook-pro
In progress
> refactor authentication to use the session cache
⏺ Read(src/auth/middleware.ts)
⏺ Update(src/auth/middleware.ts)
✻ Cooking… (4m 12s · esc to interrupt)
Codex·session-migrationoffice-desktop
Ready to review
› add an index on sessions(user_id, expires_at)
• Read schema.prisma
• Wrote prisma/migrations/0043_add_idx.sql
Hermes·server-healthcloud-server
Needs your input
› check the server’s disk usage
⟡ /var/log is using 28 GB
⟡ Archived logs account for 21 GB
Pi·nightly-backuphome-server
In progress
> check tonight’s backup and summarize the result
Reading the backup job logs…
✓ Documents copied to the home NAS
· 18m
OpenCode·auth-testscloud-server
In progress
> run the sign-in integration suite
→ bash npm run test:integration -- auth
✓ Session creation
▣ Build · deepseek-v4 · 12m
Choose a layout for 5 panes
  • Middle + sides⌘S then 1
  • Auto⌘S then 2
  • 2 columns⌘S then 3
  • 3 columns⌘S then 4
  • 4 columns⌘S then 5
  • 5 columns⌘S then 6

Agents that never sleep.

Persistent sessions. Your agents run on the machine, not in the window. Leave them on an always-on machine. Close the laptop, switch computers, drop off the Wi-Fi. Open Harness again. Same sessions. More progress.

Harness
MACHINES
Pi·nightly-backuphome-server
> check tonight’s backup and summarize the result
· 18m
Codex·session-migrationoffice-desktop
add an index on sessions(user_id, expires_at)
Read schema.prisma
Wrote prisma/migrations/0043_add_idx.sql
Ran npx prisma validate
✓ Migration ready to review
gpt-5.6-sol default · ~/work/api
OpenCode·auth-testscloud-server
> run the sign-in integration suite
▣ Build · deepseek-v4 · 12m

No SSH.
No VPN.
No open ports.

Link your machines with a remote password. Harness connects directly when your network allows it, with encrypted relay paths when it doesn’t. Every path is encrypted end to end, and only your machines hold the keys.

Direct, when your network allows it.

Your computer and the remote machine negotiate a direct WebRTC data channel. Terminal traffic travels between the two machines without a relay in the data path. The connection is still encrypted end to end.

The ultimate Harness setup.

The app brings your agents together. The Harness device puts the controls on your desk. Scroll a terminal, switch panes, or give a voice command.

Harness on a desk in front of a monitor running a terminal

Before you start.

How do remote sessions stay responsive?

Harness uses a direct WebRTC connection when your network allows it, so terminal traffic travels between your machines without a relay in the data path. It sends terminal input and output, rather than streaming a remote desktop. The actual response time still depends on your network and host machine. Your agent’s model speed is separate.

What if a direct connection is blocked?

Harness can carry the encrypted connection through Cloudflare TURN or use a WebSocket relay while WebRTC connects. It can switch to a direct connection when one becomes available. You don’t need to open inbound ports or configure a VPN.

Can a relay read my terminal traffic?

Remote terminal traffic is encrypted end to end, with the keys held by your connected machines. Relays forward the encrypted data. That protects the terminal stream; it doesn’t mean the service has no account or connection metadata.

What makes the sessions persistent?

The agent process runs on its host machine independently of the Harness window. Closing the app or losing your connection doesn’t stop work on an awake remote host. Reconnect to the same session when you’re ready.

What if the machine running an agent sleeps or restarts?

Persistent sessions still need a running host. A sleeping machine can’t keep doing work, and a reboot can stop its agent processes. For jobs you want to leave running, use an always-on home server, office machine, or cloud server.

How does Boss mode choose an agent?

It uses agent names and recent session activity across your connected machines to find a relevant session. A clear match gets your instruction. If several agents could fit, Harness asks you to choose. Routing sends work to an existing agent; it doesn’t move that agent to another machine.

Where do my agents and files actually live?

Each agent runs on the machine hosting its session and works with the files and tools available there. Putting sessions side by side in Harness doesn’t copy their projects between machines. Your agents continue to use their own model providers and credentials.

How do I connect a remote machine?

Install the Harness CLI on that machine, sign in with the same Harness account, and start it. Set a remote password, then enter it when linking the machine in the desktop app. No SSH keys to juggle.

Can I use Harness with just one computer?

Yes. Start with local agents on your Mac or Linux computer. Remote machines are optional.

Do I keep my existing agent accounts?

Yes. The agents run with your own accounts or API credentials. Harness gives you a place to work with them across your machines.

What does the free download include?

The Harness desktop app is free to download. Bring your own agent accounts; agent subscriptions, model usage, and any rented servers are separate.

Do I need the Harness device?

No. The desktop app works on its own. The device is an optional companion for controlling your sessions from your desk.

Your agents are ready. You?