One tool, many jobs
However you use it, the shape is the same: ServerMind is self-hosted (no cloud backend, no telemetry), it talks to your server through a read-only command allowlist so there's no raw shell to exploit, and any change — a restart, a stop — is refused until you flip a server-side arm switch the AI can't touch. You bring your own AI: the free Google Gemini tier, a Claude subscription, or any OpenAI-compatible API including local Ollama for zero data egress.
$ curl -fsSL https://servermind.dev/install.sh | bash