Skip to content

tunnel-manager

An SSH tunnel, remote-execution and MCP/agent server for the agent-utilities ecosystem — the agentless execution arm that lets agents run commands, transfer files and audit fleets of remote hosts over plain SSH.

Official documentation

This site is the canonical reference for tunnel-manager, maintained alongside every release.

PyPI MCP Server License GitHub

Overview

tunnel-manager provisions SSH tunnels to remote hosts and exposes them as typed, deterministic MCP tools and a Pydantic-AI agent. It is the agentless execution arm of the Agent OS — agents call it to run commands, copy files, bootstrap passwordless SSH and audit hosts at scale without deploying persistent daemons. It provides:

  • HostManager and Tunnel — a paramiko / asyncssh client layer over the inventory and individual SSH connections, with native Linux and Windows path and certificate resolution (including Teleport tsh proxy tunnelling).
  • Seven action-routed MCP tools — host inventory, single-host remote operations, bulk inventory operations, operation lifecycle, system intelligence, advanced file operations and security auditing, each toggled independently.
  • An integrated Pydantic-AI agent — exposed over the Agent Control Protocol with an optional web UI, wired to the MCP server through MCP_URL.

Explore the documentation

  • Installation — pip, source, extras, and the prebuilt Docker image.
  • Deployment — run the MCP and agent servers, Docker Compose, Caddy + Technitium.
  • Usage — the MCP tools, the HostManager / Tunnel API, and the CLI.
  • Overview — ecosystem role, distributed SSH swarm scaling, MCP configuration.
  • Teleport Architecture — certificate, proxy and cross-OS connection model.
  • Concepts — the CONCEPT:TUN-* registry.

Quick start

pip install "tunnel-manager[mcp]"
tunnel-manager-mcp                   # stdio MCP server (default transport)

Run it as a network server with a published port:

tunnel-manager-mcp --transport streamable-http --host 0.0.0.0 --port 8000

See Installation and Deployment for the full matrix (PyPI extras, Docker image, every transport, the agent server, reverse proxy and DNS).