Skip to content

Scope & Non-Goals

What OpenTunnel v1 deliberately does not do, and why that is the security model.

OpenTunnel v1 deliberately keeps the access model temporary and narrow. The exclusions below are not a roadmap of missing features; they protect the core product principle: one foreground host process, one client, one temporary CLI, and zero persistent relay state.

Every item on this list is something that would create standing access, standing state, or standing infrastructure: exactly what OpenTunnel exists to avoid.

No standing identity or access:

  • Accounts, teams, login, tokens, dashboards, or billing.
  • Install-to-system flows or daemon mode.
  • Package-manager distribution.

No standing state:

  • Persistent relay state.
  • Persistent command logs, audit logs, payload logs, or client metadata.
  • Public relay dashboard or session list.

No expanded execution surface:

  • Raw SSH compatibility.
  • PTY support.
  • Interactive stdin.
  • File upload or download.
  • Multiple simultaneous clients for one tunnel.
  • Concurrent command execution in one tunnel.
  • Background command management.

No additional integration surface:

  • MCP integration.
  • Approval workflows.

If your task needs interactive shells, file sync, or always-on access, OpenTunnel is the wrong tool; SSH and its ecosystem do that well. OpenTunnel covers the case those tools handle badly: giving an agent temporary, revocable, end-to-end encrypted command execution on a machine, with nothing to set up beforehand and nothing left behind afterwards.