putty/doc/man-pag.but

\cfg{man-identity}{pageant}{1}{2015-05-19}{PuTTY tool suite}{PuTTY tool suite}

\H{pageant-manpage} Man page for Pageant

\S{pageant-manpage-name} NAME

\cw{pageant} - PuTTY SSH authentication agent

\S{pageant-manpage-synopsis} SYNOPSIS

\c pageant ( -X | -T | --permanent | --debug ) [ key-file... ]
\e bbbbbbb   bb   bb   bbbbbbbbbbb   bbbbbbb     iiiiiiii
\c pageant [ key-file... ] --exec command [ args... ]
\e bbbbbbb   iiiiiiii      bbbbbb iiiiiii   iiii
\c pageant -a key-file...
\e bbbbbbb bb iiiiiiii
\c pageant ( -d | --public | --public-openssh ) key-identifier...
\e bbbbbbb   bb   bbbbbbbb   bbbbbbbbbbbbbbbb   iiiiiiiiiiiiii
\c pageant -D
\e bbbbbbb bb
\c pageant -l
\e bbbbbbb bb

\S{pageant-manpage-description} DESCRIPTION

\c{pageant} is both an SSH authentication agent, and also a tool for
communicating with an already-running agent.

When running as an SSH agent, it listens on a Unix-domain socket for
connections from client processes running under your user id. Clients
can load SSH private keys into the agent, or request signatures on a
given message from a key already in the agent. This permits one-touch
authentication by SSH client programs, if Pageant is holding a key
that the server they are connecting to will accept.

\c{pageant} can also act as a client program itself, communicating
with an already-running agent to add or remove keys, list the keys, or
extract their public half.

The agent protocol used by \c{pageant} is compatible with the PuTTY
tools and also with other implementations such as OpenSSH's SSH client
and \e{ssh-agent(1)}.

To run \c{pageant} as an agent, you must provide an option to tell it
what its \e{lifetime} should be. Typically you would probably want
Pageant to last for the duration of a login session, in which case you
should use either \cw{-X} or \cw{-T}, depending on whether your login
session is GUI or purely terminal-based respectively. For example, in
your X session startup script you might write

\c eval $(pageant -X)
\e bbbbbbbbbbbbbbbbbb

which will cause Pageant to start running, monitor the X server to
notice when your session terminates (and then it will terminate too),
and print on standard output some shell commands to set environment
variables that client processes will need to find the running agent.

In a terminal-based login, you could do almost exactly the same thing
but with \cw{-T}:

\c eval $(pageant -T)
\e bbbbbbbbbbbbbbbbbb

This will cause Pageant to tie its lifetime to that of your
controlling terminal: when you log out, and the terminal device ceases
to be associated with your session, Pageant will notice that it has no
controlling terminal any more, and will terminate automatically.

In either of these modes, you can also add one or more private keys as
extra command-line arguments, e.g.

\c eval $(pageant -T ~/.ssh/key.ppk)
\e bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb

in which case Pageant will prompt for the keys' passphrases (if any)
and start the agent with those keys already loaded. Passphrase prompts
will use the controlling terminal if one is available, or failing that
the GUI if one of those is available. If neither is available, no
passphrase prompting can be done.

To use Pageant to talk to an existing agent, you can add new keys
using \cw{-a}, list the current set of keys' fingerprints and comments
with \cw{-l}, extract the full public half of any key using
\cw{--public} or \cw{--public-openssh}, delete a key using \cw{-d}, or
delete all keys using \cw{-D}.

\S{pageant-manpage-lifetime} LIFETIME

The following options are called \e{lifetime modes}. They all request
Pageant to operate in agent mode; each one specifies a different
method for Pageant to start up and know when to shut down.

\dt \cw{-X}

\dd Pageant will open a connection to your X display, and when that
connection is lost, it will terminate. This gives it the same lifetime
as your GUI login session, so in this mode it is suitable for running
from a startup script such as \cw{.xsession}. The actual agent will be
a subprocess; the main Pageant process will terminate immediately,
after printing environment-variable setting commands on standard
output which should be installed in any process wanting to communicate
with the agent.

\lcont{

The usual approach would be to run

\c eval $(pageant -X)
\e bbbbbbbbbbbbbbbbbb

in an X session startup script. However, other possibilities exist,
such as directing the standard output of \cq{pageant -X} to a file
which is then sourced by any new shell.

}

\dt \cw{-T}

\dd Pageant will tie its lifetime to that of the login session running
on its controlling terminal, by noticing when it ceases to have a
controlling terminal (which will automatically happen as a side effect
of the session leader process terminating). Like \cw{-X}, Pageant will
print environment-variable commands on standard output.

\dt \cw{--exec} \e{command}

\dd Pageant will run the provided command as a subprocess, preloaded
with the appropriate environment variables to access the agent it
starts up. When the subprocess terminates, Pageant will terminate as
well.

\lcont{

All arguments on Pageant's command line after \cw{--exec} will be
treated as part of the command to run, even if they look like other
valid Pageant options or key files.

}

\dt \cw{--permanent}

\dd Pageant will fork off a subprocess to be the agent, and print
environment-variable commands on standard output, like \cw{-X} and
\cw{-T}. However, in this case, it will make no effort to limit its
lifetime in any way; it will simply run permanently, unless manually
killed. The environment variable \cw{SSH_AGENT_PID}, set by the
commands printed by Pageant, permits the agent process to be found for
this purpose.

\lcont{

This option is not recommended, because any method of manually killing
the agent carries the risk of the session terminating unexpectedly
before it manages to happen.

}

\dt \cw{--debug}

\dd Pageant will run in the foreground, without forking. It will print
its environment variable setup commands on standard output, and then it
will log all agent activity to standard output as well. This is useful
for debugging what Pageant itself is doing, or what another process is
doing to it.

\S{pageant-manpage-client} CLIENT OPTIONS

The following options tell Pageant to operate in client mode,
contacting an existing agent via environment variables that it should
already have set.

\dt \cw{-a} \e{key-files}

\dd Load the specified private key file(s), decrypt them if necessary
by prompting for their passphrases, and add them to the
already-running agent.

\lcont{

The private key files must be in PuTTY's \cw{.ppk} file format.

}

\dt \cw{-l}

\dd List the keys currently in the running agent. Each key's
fingerprint and comment string will be shown.

\dt \cw{--public} \e{key-identifiers}

\dd Print the public half of each specified key, in the RFC 4716
standard format (multiple lines, starting with \cq{---- BEGIN SSH2
PUBLIC KEY ----}).

\lcont{

Each \e{key-identifier} can be any of the following:

\b The name of a file containing the key, either the whole key (again
in \cw{.ppk} format) or just its public half.

\b The key's comment string, as shown by \cw{pageant -l}.

\b Enough hex digits of the key's fingerprint to be unique among keys
currently loaded into the agent.

If Pageant can uniquely identify one key by interpreting the
\e{key-identifier} in any of these ways, it will assume that key was
the one you meant. If it cannot, you will have to specify more detail.

If you find that your desired \e{key-identifier} string can be validly
interpreted as more than one of the above \e{kinds} of identification,
you can disambiguate by prefixing it with \cq{file:}, \cq{comment:} or
\cq{fp:} to indicate that it is a filename, comment string or
fingerprint prefix respectively.

}

\dt \cw{--public-openssh} \e{key-identifiers}

\dd Print the public half of each specified key, in the one-line
format used by OpenSSH, suitable for putting in
\cw{.ssh/authorized_keys} files.

\dt \cw{-d} \e{key-identifiers}

\dd Delete each specified key from the agent's memory, so that the
agent will no longer serve it to clients unless it is loaded in again
using \cw{pageant -a}.

\dt \cw{-D}

\dd Delete all keys from the agent's memory, leaving it completely
empty.

\S{pageant-manpage-options} OPTIONS

\dt \cw{-v}

\dd Verbose mode. When Pageant runs in agent mode, this option causes
it to log all agent activity to its standard error. For example, you
might run

\lcont{

\c eval $(pageant -X -v 2>~/.pageant.log)
\e bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb

and expect a list of all signatures requested by agent clients to
build up in that log file.

The log information is the same as that produced by the \cw{--debug}
lifetime option, but \cw{--debug} sends it to standard output (since
that is the main point of debugging mode) whereas \cw{-v} in all other
lifetime modes sends the same log data to standard error (being a
by-product of the program's main purpose). Using \cw{-v} in
\cw{--debug} mode has no effect: the log still goes to standard
output.

}

\dt \cw{-s}, \cw{-c}

\dd Force Pageant to output its environment setup commands in the
style of POSIX / Bourne shells (\cw{-s}) or C shells (\cw{-c})
respectively. If neither option is given, Pageant will guess based on
whether the environment variable \cw{SHELL} has a value ending in
\cq{csh}.

\dt \cw{--help}

\dd Print a brief summary of command-line options and terminate.

\dt \cw{--version}

\dd Print the version of Pageant.

\dt \cw{--}

\dd Cause all subsequent arguments to be treated as key file names,
even if they look like options.