TurboTunnel creates on-demand ssh tunnels. It listens on local ports, starts up ssh connections when something connects to those ports, and proxies data through the remote tunnel.

Trust me, it's magic.

$ go get github.com/zwily/turbotunnel
$ go build github.com/zwily/turbotunnel

tunnels:
  - name: Work Intranet
    localPort: 10001
    jumpHost: jump1.example.com
    remoteHost: 10.0.13.10
    remotePort: 80
  - name: Work Active Directory RDP
    localPort: 10002
    jumpHost: [email protected]
    remoteHost: 10.0.0.4
    remotePort: 3389

$ ./turbotunnel -config /path/to/config.yml

Once TurboTunnel is running, you can then open http://localhost:10001 in your browser. TurboTunnel will see the connection to port 10001 and initiate an ssh connection to jump1.example.com forwarding a local port to 10.0.13.10:80. TurboTunnel will then proxy all data between the opened connection and the local tunnel.

If you're using OS X, you can get desktop notifications when a tunnel is started by installing terminal-notifier:

$ brew install terminal-notifier

I run TurboTunnel via launchd. Sample configuration, placed in ~/Library/LaunchAgents/com.github.zwily.turbotunnel.plist:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>Label</key>
    <string>com.github.zwily.turbotunnel</string>
    <key>ProgramArguments</key>
    <array>
        <string>/usr/local/bin/turbotunnel</string>
        <string>-config</string>
        <string>/Users/zach/.turbotunnel.yml</string>
    </array>
    <key>KeepAlive</key>
    <true/>
    <key>RunAtLoad</key>
    <true/>
</dict>
</plist>

To start it after putting the file in place:

$ launchctl load ~/Library/LaunchAgents/com.github.zwily.turbotunnel.plist