Open Source Connector · MIT License · Live transparency

Your AI terminal.
Watching the cockpit.

Run Claude Code, Codex, Gemini, Ollama and more from any browser — every command logged, every honeytoken armed, every alert visible to you. When AI pilots AI, you stay in the loop.

Get started free Learn more
aiterm.io/app
Machines
production-web
C Claude Code Start
O Ollama Start
dev-server
C Claude Code Start
claude ~/project $ Help me refactor the authentication module
 
I'll analyze the auth module and suggest improvements.
Looking at src/auth/...
 
Found 3 areas to improve:
1. Session handling can be simplified
2. Password hashing should use argon2
3. Token refresh logic has a race condition
 
Let me start with the session handling...
Voice · Upload AITerm EN
Features

Everything you need

A lightweight connector, a powerful platform.

⌨

Multi-AI Support

Run Claude, Ollama, llama.cpp, LM Studio, vLLM and more on a single machine — manage them all from one dashboard.

⇄

Code from anywhere

Phone, iPad, laptop. The connector reaches out to the hub — no SSH, no port forwarding, no VPN. Works on cafe Wi-Fi and mobile data.

🔑

Smart Pairing

One command to install. Click a link to pair. No tokens to copy — fully automatic.

💻

Real Terminal

xterm.js with themes, copy/paste, drag & drop uploads, voice input and touch support.

🌐

Multi-Tenant

Multiple users, multiple machines. Everyone sees only their own terminals — securely isolated.

🔌

Open Source Connector

MIT licensed. Full transparency — read the code before you install it.

Guard Mode

A safety net between AI and your shell

Per-machine toggle. Dangerous bash commands pause for your approval — with the exact command shown in front of you.

Guard Mode
Confirm dangerous command?
Detected: recursive forced delete
$ rm -rf /var/lib/mysql

What's caught

Recursive deletes, dd to block devices, curl|bash, shutdown, service stop, firewall flush, fork bombs, authorized_keys overwrite.

What's not

Ordinary commands fly through untouched. You are not asked about git push, ls, or AI sessions — only bash.

Transparent

Pattern list is open source. Security through user-in-loop, not obscurity.

Security

Built for teams that read the code

Every layer — installer, updates, sessions — is designed to be auditable.

🔒

Signed installer & updates

Ed25519-signed manifest with embedded public key. Every downloaded file is hash-verified before execution — even on first install.

🔗

MIT-licensed connector

Full source on GitHub. Read it before you install it. Fork it if you want. The hub stays closed; the agent on your machine does not.

🛡

Responsible disclosure

Safe Harbor language. Report via email or GitHub Private Vulnerability Reporting. SLAs documented, Hall of Fame open.

🔍

Continuous scanning

Every commit runs CodeQL, Bandit, pip-audit, gitleaks, ShellCheck. Dependency CVEs cannot reach main.

Full security policy →

Installation

Connected in 3 steps

From first command to working terminal in under 2 minutes.

1

Install

One command installs the connector and shows a pairing link.

curl -sL https://aiterm.io/install | bash
2

Pair

Open the link in your browser, sign in and confirm.

3

Use

Machine appears in the dashboard. Scan for AI backends, open terminals — all from your browser.

Ready to get started?

Connect your first machine in under 2 minutes.

Create free account
curl -sL https://aiterm.io/install | bash