mcpField
T
tunnels.io

tunnels.io

Local
tunnels.ioCloud & Infra0Updated 5w ago

About tunnels.io

tunnels-mcp is a local MCP server for reading a tunnels.io account and opening or closing tunnels. It provides account tools according to the permissions granted to its token, while keeping plan changes and deployments out of scope.

Key features

  1. 01Read account and usage details
  2. 02List tunnels, tokens, invoices, and domains
  3. 03View metrics and notifications
  4. 04Start and stop HTTP or TCP tunnels
  5. 05Expose tools allowed by token permissions
  6. 06Redact token-related error bodies

Use cases

  1. 01Inspect account usage and metrics
  2. 02Review invoices and registered domains
  3. 03Expose a local service through a tunnel
  4. 04Stop an active tunnel

MCP support

The server exposes tunnels.io account and tunnel tools to MCP clients. It lists only tools permitted by the configured token; tunnel start and stop operations use the tunnels CLI.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository
Packages
mcpbtunnels-mcp-0.0.2.mcpb0.0.2

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

tunnels-mcp

MCP server for tunnels.io.

It runs locally, holds one CLI token, and gives an assistant read access to the account. It can also open a tunnel.

Install

Download a binary for your platform from the releases page, or build it:

go install github.com/tunnels-io/tunnels-mcp/cmd/[email protected]

Clients that take MCP bundles, such as Claude Desktop, can install tunnels-mcp-<version>.mcpb from the same page. It asks for the token.

The server is listed in the MCP Registry as io.github.tunnels-io/tunnels-mcp.

Token

Create one at tunnels.io/account/tokens. Tick the permissions you want.

Skipping a permission is not a broken tool. The server reads the token's permissions at startup and lists only the tools that work.

An older all-access token (*) works for everything except billing, domains and cloud apps. Those three services refuse *, so the server does not list their tools for such a token. Make a token on the page above to reach them.

TUNNELS_TOKEN     required, starts tnl_
TUNNELS_API       default https://tunnels.io
TUNNELS_CLI       default "tunnels" on PATH
TUNNELS_INSPECT   unset. The CLI inspector is off; set an address to enable it.

A config file works too, at ~/.config/tunnels/mcp.json:

{ "token": "tnl_..." }

The environment wins.

Client setup

{
  "mcpServers": {
    "tunnels": {
      "command": "tunnels-mcp",
      "env": { "TUNNELS_TOKEN": "tnl_..." }
    }
  }
}

Tools

ToolPermission
list_tunnelstunnels:read
get_usageusage:read
get_metricsusage:read
get_accountaccount:read
list_tokenstokens:read
list_invoicesbilling:read
list_domainsregistrar:read
list_notificationsnotifications:read
list_cloud_appscloud:read
start_tunnelnone
stop_tunnelnone

Reads only, apart from the last two. Those shell out to the tunnels CLI, which must be on PATH. No tool changes a plan, buys a domain or deploys anything.

start_tunnel takes http (the default, served as HTTPS on the public URL) or tcp.

Token handling

The token goes to tunnels.io in a Bearer header, and to the CLI through the environment. Never a command line. Never a log. Never a tool result.

Error bodies are redacted before they are shown.

The server contacts tunnels.io and nothing else.

Build

make check
make build

No dependencies.

Release

Push a tag such as v0.0.2. The release workflow builds six binaries and the MCPB bundle, creates the GitHub release, and publishes server.json to the MCP Registry. To build the same files locally, on macOS:

make release V=0.0.2

Licence

MIT. See LICENSE.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.