Product
Agentic operations, end to end.
Not autocomplete. Agentic SSH plans, runs, reads the output, and decides the next step — on the servers you point it at, within the limits you set.
What’s inside
An agentic loop
It plans, executes, observes the output, and decides what to do next — instead of just suggesting one command at a time.
Three safety modes
Auto, Confirm, and Confirm-on-write. Destructive and write operations are gated by default.
Bring your own backend
Claude, OpenAI, Gemini for the smartest answers — or Ollama to keep everything local. You hold the keys.
Audit transcript
Every command, output and decision is appended to a structured JSONL log you can export and diff.
Zero target footprint
It reuses your OpenSSH config, ProxyJump and keys. Nothing to install on the server, no ports to open.
Cross-platform & signed
Native builds for macOS (signed & notarized), Windows and Linux. Localized in English, 繁中 and 简中.
See it on your own servers.
Free to start, on a single host with local models.