2 [[meta title="Documentation"]]
6 Monkeysphere relies on:
8 * [GnuTLS](http://gnutls.org/) version 2.4.0 or later
9 * [OpenSSH](http://openssh.com/)
10 * [GnuPG](http://gnupg.org/)
14 * Getting started as a [user](/getting-started-user)
15 * Getting started as a [server admin](/getting-started-admin)
19 * [Initial specifications at CMRG](http://cmrg.fifthhorseman.net/wiki/OpenPGPandSSH)
20 * [OpenPGP (RFC 4880)](http://tools.ietf.org/html/rfc4880)
21 * [Secure Shell Authentication Protocol (RFC 4252)](http://tools.ietf.org/html/rfc4252)
22 * [URI scheme for SSH, RFC draft](http://tools.ietf.org/wg/secsh/draft-ietf-secsh-scp-sftp-ssh-uri/)
26 The monkeysphere isn't the only project intending to implement a PKI
27 for OpenSSH. We provide links to these other projects because they're
28 interesting, though we have concerns with their approaches.
30 All of the other projects we've found so far require a patched version
31 of OpenSSH, which makes adoption more difficult. Most people don't
32 build their own software, and simply overlaying a patched binary is
33 associated with significant maintenance (and therefore security)
36 While ultimately contributing a patch to
37 [OpenSSH](http://openssh.com/) (or any
38 [free](http://www.chiark.greenend.org.uk/~sgtatham/putty/)
39 [SSH](http://www.lysator.liu.se/~nisse/lsh/)
40 [implementation](http://matt.ucc.asn.au/dropbear/dropbear.html)) is
41 not a bad thing, we hope to be able to better establish the use of a
42 PKI without resorting to source modification.
46 [openssh-gpg](http://www.red-bean.com/~nemo/openssh-gpg/) is a patch
47 against OpenSSH to support OpenPGP certificates. According to its
48 documentation, it is intended to support [`pgp-sign-rsa` and
49 `pgp-sign-dss` public key algorithms for hosts, as specified by the
50 IETF](http://tools.ietf.org/html/rfc4253#section-6.6).
52 Some concerns with `openssh-gpg`:
54 * This patch is old; it doesn't appear to have been maintained beyond
55 OpenSSH 3.6p1. As of this writing, OpenSSH 5.1p1 is current.
57 * It only provides infrastructure in one direction: the user
58 authenticating the host by name. There doesn't seem to be a
59 mechanism for dealing with identifying users by name, or allowing
60 users to globally revoke or update keys.
62 * The choice of User ID (`anything goes here (and here!)
63 <ssh@foo.example.net>`) for host keys overlaps with the current use
64 of the User ID space. While it's unlikely that someone actually
65 uses this e-mail address in the web of trust, it would be a nasty
66 collision, as the holder of that key could impersonate the server
67 in question. The monkeysphere uses [User IDs of the form
68 `ssh://foo.example.net`](http://tools.ietf.org/wg/secsh/draft-ietf-secsh-scp-sftp-ssh-uri/)
69 to avoid collisions with existing use.
71 * It's not clear that `openssh-gpg` acknowledges or respects the
72 [usage flags](http://tools.ietf.org/html/rfc4880#section-5.2.3.21)
73 on the host keys. This means that it could accept a "sign-only"
74 key as suitable for authenticating a host, despite the
75 clearly-marked intentions of the key-holder.
77 ### Perspectives OpenSSH client ###
79 [The Perspectives project](http://www.cs.cmu.edu/~perspectives/) at
80 CMU has released an [openssh client that uses network
81 notaries](http://www.cs.cmu.edu/~perspectives/openssh.html) to bolster
82 your confidence in newly-seen keys. This offers a defense against a
83 narrow MITM attack (e.g. by someone who controls your local gateway)
84 by simply verifying that other machines from around the network see
85 the same keys for the remote host that you're seeing.
87 This tactic is quite useful, but doesn't take the system as far as it
88 could go, and doesn't tie into any existing web of trust.
90 Some concerns with the Perspectives OpenSSH client:
92 * This client won't help if you are connecting to machines behind
93 firewalls, on NAT'ed LANs, with source IP filtering, or otherwise
94 in a restricted network state.
96 * There is still a question of why you should trust these particular
97 notaries during your verification. Who are the notaries? How
98 could they be compromised?
100 * It only provides infrastructure in one direction: the user
101 authenticating the host by name. There is no mechanism for dealing
102 with identifying users by name, or allowing users to globally
103 revoke or change keys.
105 * It doesn't provide any mechanism for key rotation or revocation:
106 Perspectives won't help you if you need to re-key your machine.
108 ### OpenSSH with X.509v3 certificates ###
110 Roumen Petrov [maintains a patch to OpenSSH that works with the X.509
111 PKI model](http://www.roumenpetrov.info/openssh/). This is the
112 certificate hierarchy commonly used by TLS (and SSL).
114 Some concerns about OpenSSH with X.509v3:
116 * the X.509 certificate specification itself [encourages corporate
117 consolidation and centralized global "trust" because of its
118 single-issuer architectural
119 limitation](http://lair.fifthhorseman.net/~dkg/tls-centralization/).
120 This results in an expensive and cumbersome system for smaller
121 players, and it also doesn't correspond to the true distributed
122 nature of human-to-human trust. Furthermore, centralized global
123 "trusted authorities" create a tempting target for attack, and a
124 single-point-of-failure if an attack is successful.
126 Depending on how you declare your trust relationships, OpenPGP is
127 capable of providing the same hierarchical structure as X.509, but
128 it is not limited to such a structure. The OpenPGP Web of Trust
129 model is more flexible and more adaptable to represent real-world
130 trust than X.509's rigid hierarchy.
132 * X.509 certificates can identify hosts by name, but not by
133 individual service. This means that a compromised web or e-mail
134 server with access to the X.509 key for that service could re-use
135 its certificate as an SSH server, and it would be able to
136 masquerade successfully.
138 The monkeysphere uses [User IDs of the form
139 `ssh://foo.example.net`](http://tools.ietf.org/wg/secsh/draft-ietf-secsh-scp-sftp-ssh-uri/),
140 so they are not by-default shared across services on the same host
141 (you can still share a key across services on the same host if you
142 like, but the service User IDs can be certified independently of