trezor-agent/doc/README-SSH.md

SSH Agent

1. Configuration

SSH requires no configuration, but you may put common command line options in ~/.ssh/agent.conf to avoid repeating them in every invocation.

See (trezor|keepkey|ledger)-agent -h for details on supported options and the configuration file format.

If you'd like a Trezor-style PIN entry program, follow these instructions.

2. Usage

Use the (trezor|keepkey|ledger)-agent program to work with SSH. It has three main modes of operation:

1. Export public keys

To get your public key so you can add it to authorized_hosts or allow ssh access to a service that supports it, run:

(trezor|keepkey|ledger)-agent identity@myhost

The identity (ex: identity@myhost) is used to derive the public key and is added as a comment to the exported key string.

2. Run a command with the agent's environment

Run

$ (trezor|keepkey|ledger)-agent identity@myhost -- COMMAND --WITH --ARGUMENTS

to start the agent in the background and execute the command with environment variables set up to use the SSH agent. The specified identity is used for all SSH connections. The agent will exit after the command completes.

As a shortcut you can run

$ (trezor|keepkey|ledger)-agent identity@myhost -s

to start a shell with the proper environment.

2. Connect to a server directly via (trezor|keepkey|ledger)-agent

If you just want to connect to a server this is the simplest way to do it:

$ (trezor|keepkey|ledger)-agent user@remotehost -c

The identity user@remotehost is used as both the destination user and host as well as for key derivation, so you must generate a separate key for each host you connect to.

3. Common Use Cases

Start a single SSH session

Start multiple SSH sessions from a sub-shell

This feature allows using regular SSH-related commands within a subprocess running user's shell. SSH_AUTH_SOCK environment variable is defined for the subprocess (pointing to the SSH agent, running as a parent process). This way the user can use SSH-related commands (e.g. ssh, ssh-add, sshfs, git, hg), while authenticating via the hardware device.

Load different SSH identities from configuration file

Implement passwordless login

Run:

/tmp $ trezor-agent user@ssh.hostname.com -v > hostname.pub
2015-09-02 15:03:18,929 INFO         getting "ssh://user@ssh.hostname.com" public key from Trezor...
2015-09-02 15:03:23,342 INFO         disconnected from Trezor
/tmp $ cat hostname.pub
ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBGSevcDwmT+QaZPUEWUUjTeZRBICChxMKuJ7dRpBSF8+qt+8S1GBK5Zj8Xicc8SHG/SE/EXKUL2UU3kcUzE7ADQ= ssh://user@ssh.hostname.com

Append hostname.pub contents to /home/user/.ssh/authorized_keys configuration file at ssh.hostname.com, so the remote server would allow you to login using the corresponding private key signature.

Access remote Git/Mercurial repositories

Copy your public key and register it in your repository web interface (e.g. GitHub):

$ trezor-agent -v -e ed25519 git@github.com | xclip

Use the following Bash alias for convenient Git operations:

$ alias git_hub='trezor-agent -v -e ed25519 git@github.com -- git'

Replace git with git_hub for remote operations:

$ git_hub push origin master

The same works for Mercurial (e.g. on BitBucket):

$ trezor-agent -v -e ed25519 git@bitbucket.org -- hg push

Start the agent as a systemd unit

1. Create these files in ~/.config/systemd/user

Replace trezor with keepkey or ledger as required.

trezor-ssh-agent.service

[Unit]
Description=trezor-agent SSH agent
Requires=trezor-ssh-agent.socket

[Service]
Type=Simple
Environment="DISPLAY=:0"
Environment="PATH=/bin:/usr/bin:/usr/local/bin:%h/.local/bin"
ExecStart=/usr/bin/trezor-agent --foreground --sock-path %t/trezor-agent/S.ssh IDENTITY

If you've installed trezor-agent locally you may have to change the path in ExecStart=.

Replace IDENTITY with the identity you used when exporting the public key.

trezor-ssh-agent.socket

[Unit]
Description=trezor-agent SSH agent socket

[Socket]
ListenStream=%t/trezor-agent/S.ssh
FileDescriptorName=ssh
Service=trezor-ssh-agent.service
SocketMode=0600
DirectoryMode=0700

[Install]
WantedBy=sockets.target

2. Run

systemctl --user start trezor-ssh-agent.service trezor-ssh-agent.socket
systemctl --user enable trezor-ssh-agent.socket

3. Add this line to your .bashrc or equivalent file:

export SSH_AUTH_SOCK=$(systemctl show --user --property=Listen trezor-ssh-agent.socket | grep -o "/run.*")

4. SSH will now automatically use your device key in all terminals.

4. Troubleshooting

If SSH connection fails to work, please open an issue with a verbose log attached (by running trezor-agent -vv) .

Incompatible SSH options

Note that your local SSH configuration may ignore trezor-agent, if it has IdentitiesOnly option set to yes.

 IdentitiesOnly
         Specifies that ssh(1) should only use the authentication identity files configured in
         the ssh_config files, even if ssh-agent(1) or a PKCS11Provider offers more identities.
         The argument to this keyword must be “yes” or “no”.
         This option is intended for situations where ssh-agent offers many different identities.
         The default is “no”.

If you are failing to connect, try running:

$ trezor-agent -vv user@host -- ssh -vv -oIdentitiesOnly=no user@host