Hey Product Hunt 👋 I'm the founder of Termalin. I built it because I got nervous watching AI coding agents SSH into real servers — the usual advice is "hand the agent a private key and a raw shell and hope." That felt wrong for anything touching production. So Termalin is first a proper SSH client I actually keep open all day — tabbed terminal, SFTP with an in-place editor, a dual-pane file manager, port forwarding, a live system-stats bar, E2E-encrypted sync. Rust core, so it stays light. Windows, macOS and Linux. The part no other SSH client has: a built-in MCP server. Point Claude Code (or any MCP client) at it and: • Termalin acts as a key custodian — the agent and the model never see a private key. • Access is off by default. You choose which hosts, and set each to full / an allowlist / blocked. • You watch every session live in a "watch grid" — agent-driven tiles glow — and take the keyboard any time. • Everything (host connections, file ops, agent commands) lands in an audit log. Recording captures output only, never keystrokes. Being straight about where it's at: • Desktop only for now — Windows/macOS/Linux. Mobile is planned; today there's a browser cabinet for phones. • No split panes / command palette / rebindable keys yet — those are on the list, tell me if they matter to you. • Already on the Microsoft Store, with a signed one-click install on Windows. Free tier has no host limits, and every new account gets a 14-day Pro trial (no card). I'll be around all day — genuinely want the hard questions, especially on the agent-safety model. What would you need to trust an agent on your own servers? — Svyatoslav

Termalin

Safe SSH for you and your AI agents

Safe SSH for you and your AI agents